Reviewer #1 (Public Review):

In this manuscript, Blanc et al. developed a tool to align different larval zebrafish brains with pan-neural markers and additional birthdate labeling into a common atlas. By aligning transgenic lines into this reference atlas, the authors tried to infer the birth date and growth patterns of different neuron populations. The intention of providing an open-access tool and developmental atlas is good, especially considering most of the current zebrafish brain atlases were made for adult or larval zebrafish more than 5 days old. However, the key features claimed by the authors i.e., the "temporal dynamic" is essentially missing from the atlas. The tool was still built for a single development stage and reflected no information on growth patterns except the neuronal birthdate. Moreover, the accuracy of the registration method, the rationality of the birthdate labeling, and the validity of the proof-of-concept inference were also not sufficiently demonstrated in the experimental design.

Overall, I believe the manuscript has the potential to be a useful tool and an impactful developmental atlas for the community, but it would need substantial improvement in method design, experimental validation, and data/software availability.

Major points:

1. The authors claimed to have made a "3D-temporal" atlas for developing zebrafish hindbrain. However, the "temporal" component was solely birthdate inferred from temporal labeling. Images were still acquired at the same developmental stage, which makes the atlas and registration method not substantially different from the other existing atlases (e.g. ViBE-Z (Ronneberger 2012), Z-Brain (Randlett 2015), ZBB (Tabor 2019), Mapzebrain (Kunst 2019) - note not all of these tools were cited in introduction). The authors would have to either add temporal tracing of the population and provide registration between different developmental stages, or tune down the "temporal" term only to "birthdating".

2. Rigid registration was used to align the images from different individuals, as opposed to the more complicated non-linear registration used by all the tools above. The accuracy of such registration needs to be measured to justify the choice of method, by measuring the inter-individual variability using different registration methods. Variability should be quantified in 3D rather than along specific anatomical axes.

3. Birthdate labeling was achieved by photoconverting Kaede at different stages (24, 36, 48 hpf) and imaging at 72hpf. This method suffers from an intrinsic bias: the Kaede-red was subject to different time windows for diluting and metabolizing over development, making the age labeling incomparable between different labeling lengths. To verify the experimental design, the authors should 1) demonstrate that the red cells labeled in an early conversion are strictly included in the red cells labeled in a late conversion, and 2) provide an additional age-labeling method like BrdU treatment, to show the new cells incorporated between the two time points are reflected in the growing photoconverted population.

4. Proof-of-concept inference of GABAergic neuron birth date in Figure 5 is very vague. No link was shown between the red cells in Fig 5B and the gad1b in situ-positive cells in Fig 5D. If tracing the fate of these cells from 24-72hpf is not possible, the authors should at least demonstrate that they are 1) post-mitotic at 24hpf, i.e. HuC-positive; and 2) appear in similar numbers and similar neighborhood context as the red cells in Fig 5B. I also want to point out that while it is true that mRNAs are expressed earlier than fluorescent proteins in the transgenic line, an early-born cell expressing a specific gene late development does not mean it would express the gene early on. A gene can be ON early on and turned OFF later; Conversely, a gene can express late in the differentiation process while the cell is committed and went through terminal division early in the lineage.

5. It is mentioned many times that the platform is "open-access" and "expandable", but no source or browsable atlas was provided (maybe I was wrong, but I did not find the Fiji macro and R code on the provided website). The software and data availability should be improved, and more demonstration is needed to show its "expendability" -guidelines should be provided on how to upload users' own data to use this platform, and what kind of additional data is supported.

Author Response

Reviewer 2

The manuscript by Huisjes et al presented an open-source platform for the storage and processing of imaging data, particularly for single-molecule imaging experiments. Compared to sequencing data, which have a more standardized format for data storage, imaging data have more diverse formats due to the fact that different research labs tend to use different instruments and software (either commercial or home-built) for data collection and analysis. Manual input is almost always necessary at certain steps of data analysis. All these create difficulties in data storage and reproducibility. The authors provide a practical solution to this problem by the molecular archive suite, "Mars". This platform is integrated into imageJ/Fiji, and can be used for storing detailed description of experimental settings, performing standard imaging processing steps, and recording manual input information during data analysis. I judge this platform, if fully functional and generalizable, will be very useful to many labs who are using single-molecule imaging methods in the research.

Strength:

1. The work presented a fairly user friendly interface (using Fiji directly), and fairly detailed protocol and other documentations in a very nicely designed website. I was able to download and use it based on the tutorial.

2. It is integrated very well with Fiji, and some analysis modules are directly from existing Fiji analysis/plugins.

Weakness:

I invited one of my students to co-test the suite. We tried on both Mac and Windows systems, using the example FRET data set described in the manuscript and one of our own single-molecule images. We encountered some technical issues.

We are very happy with the overall positive assessment of the reviewer that Mars could offer a common format that helps to enforce reproducible analysis workflows that can easily be shared with others.

We are grateful for the additional feedback and testing done by the reviewer and her student. Ensuring that Mars works as expected on all computers and configurations is difficult given that we don’t have them at hand for testing ourselves. During the revision period, we have done more testing on more computer systems and we hope we have addressed the issues. We believe it will be impossible for us to guarantee that Mars works without problems on the first try for everyone. Therefore, Mars is a community partner on the Scientific Community Image Forum where users can report their problems in posts with the mars tag and we can help troubleshoot them (https://forum.image.sc/tag/mars). We believe this approach will offer the best support going forward. Nevertheless, we continue to make improvements and test to make sure all bugs we discover are addressed.

In the revision, we completely reworked the smFRET example workflow and added two additional workflows to address all the comments from the reviewers and reviewing editor. In addition to expanding the explanations, and troubleshooting information on the Mars documentation website, we also created a YouTube channel with tutorial and example videos (https://www.youtube.com/channel/UCkkYodMAeotj0aYxjw87pBQ). We go through the new dynamic smFRET workflow from start to finish in one of the videos provided (https://www.youtube.com/watch?v=JsyznI8APlQ). We hope this will make it clear what inputs and outputs are expected and how the workflow should proceed. This was done on a mac but we have also tested this workflow on windows without encountering problems.

Emissionen durch die Verfeuerung der vorhandenen Assets: 650 Gigatonnen

Das bedeutet eine 30prozentige Überschreitung des CO2-Budgets für 50% Wahrscheinlichkeit des 1,5°-Ziels

and seriously you don't mention visual code???

This is the longest sentence in The Garden Party according to our last homework. It uses em dashes and long, descriptive phrases with commas to compare and contrast Leila's school balls and her first real ball.

The author uses personification to describe the surroundings around the protagonist, which gave me the idea in creating some program that analyzes the use of personification within literature and determine how accurate the findings are.

If you would like to update pip on Windows, use python -m pip install --upgrade pip

These maps are the most common map types that I have seen. These are used in tourist maps, college building maps, museum maps, and such. I have also used this kind of map using the Google My Map to relay information about Japanese American incarceration on the west coast, pointing out specific locations with text, website info, color, to show the development of Japanese towns.

Author Response:

Reviewer #3 (Public Review):

Murphy et al. further develop the linked selection model of Elyashiv et al. (2016) and apply it to human genetic variation data. This model is itself an extension of the McVicker et al. (2009) paper, which developed a statistical inference method around classic background selection (BGS) theory (Hudson and Kaplan, 1995, Nordborg et al., 1996). These methods fit a composite likelihood model to diversity data along the chromosome, where the level of diversity is reduced by a local factor from some initial "neutral" level π0 down to observed levels. The level of reduction is determined by a combination of both BGS and the expected reduction around substitutions due to a sweep (though the authors state that these models are robust to partial and soft sweeps). The expected reduction factor is a function of local recombination rates and genomic annotation (such as exonic and phylogenetically conserved sequences), as well as the selection parameters (i.e. mutation rates and selection coefficients for different annotation classes). Overall, this work is a nice addition to an important line of work using models of linked selection to differentiate selection processes. The authors find that positive selection around substitutions explains little of the variation in diversity levels across the genome, whereas a background selection model can explain up to 80% of the variance in diversity. Additionally, their model seems to have solved a mystery of the McVicker et al. (2009) paper: why the estimated deleterious mutation rate was unreasonably high. Throughout the paper, the authors are careful not only in their methodology but also in their interpretation of the results. For example, when interpreting the good fit of the BGS model, the authors correctly point out that stabilizing selection on a polygenic trait can also lead to BGS-like reductions.

Furthermore, the authors have carefully chosen their model's exogenous parameters to avoid circularity. The concern here is that if the input data into the model - in particular the recombination maps and segments liked to be conserved - are estimated or identified using signals in genetic variation, the model's good fit to diversity may be spurious. For example, often recombination maps are estimated from linkage disequilibrium (LD) data which is itself obtained from variation along the chromosome. Murphy et al. use a recombination map based on ancestry switches in African Americans which should prevent "information leakage" between the recombination map and the BGS model from leading to spuriously good fits. Likewise, the authors use phylogenetic conservation maps rather than those estimated from diversity reductions (such as McVicker et al.'s B maps) to avoid circularity between the conserved annotation track and diversity levels being modeled. Additionally, the authors have carefully assessed and modified the original McVicker et al. algorithm, reducing relative error (Figure A2).

One could raise the concern that non-equilibrium demography confounds their results, but the authors have a very nice analysis in Section 7 of the supplementary material showing that their estimates are remarkably stable when the model is fit separately in different human populations (Figure A35). Supporting previous work that emphasizes the dependence between BGS and demography, the authors find evidence of such an interaction with a clever decomposition of variance approach (Figure A37). The consistency of BGS estimates across populations (e.g. Figures A35 and A36) is an additional strong bit of evidence that BGS is indeed shaping patterns of diversity; readers would benefit if some of these results were discussed in the main text.

We appreciate the reviewer’s kind remarks. With regards to the results included in the main text vs the supplement, we attempted to strike a balance between having the main text remain communicative to a larger readership and providing experts with details they may find useful. We have, however, done our best for the supplementary analyses to be written clearly.

I have three major concerns about this work. First, it's unclear how accurate the selection coefficient estimates are given the non-equilibrium demography of humans (pre-Out of Africa split, and thus not addressed by the separate population analyses). The authors do not make a big point about the selection coefficient estimates in the main section of the paper, so I don't find this to be a big problem. Still, some mention of this issue might be helpful to readers trying to interpret the results presented in the supplementary text.

As the reviewer notes, we chose not to emphasize the inferred distributions of selection coefficients. Our main reason for this choice is the technical issue addressed in Appendix Section 1.5 (L561-564): “Second, thresholding potentially biases our estimates of the distribution of selection effects. While this bias is probably smaller than the bias without thresholding, its form and magnitude are not obvious. This is why we decided not to report the inferred distributions of selection effects in the Main Text.” We agree that if we were to focus on our estimates of the distribution of selection effects, the effects of demographic history would also need to be considered. This is, however, not the focus here.

Second, I'm curious whether the composite likelihood BGS model could overfit any variance along the chromosome - even neutral variance. At some level, the composite likelihood approach may behave like a sort of smoothing algorithm, albeit with a functional form and parameters of a BGS model. The fact that there is information sharing across different regions with the same annotation class should in principle prevent overfitting to local noise. Still, there are two ways I think to address this overfitting concern. First, a negative neutral control could help - how much variation in diversity along the chromosome can this model explain in a purely neutral simulation? I imagine very little, likely less than 5%, but I think this paper would be much stronger with the addition of a negative control like this. Second, I think the main text should include the R2 values from out-sample predictions, rather than just the R2 estimates from the model fit on the entire data. For example, one could fit the model on 20 chromosomes, use the estimated θΒ parameters to predict variation on the remaining two. The authors do a sort of leave-one-out validation at the window level (Figure A31); however, this may not be robust to linkage disequilibrium between adjacent windows in the way leaving out an entire chromosome would be.

The two requested analyses were done and their results are described above, in response to essential revisions (p. 2-3 here). In brief, there is no overfitting of neutral patterns or otherwise. We elaborate on why this finding is expected below.

Finally, I feel like this paper would be stronger with realistic forward simulations. The deterministic simulations described in the supplementary materials show the implementation of the model is correct, but it's an exact simulation under the model - and thus not testing the accuracy of the model itself against realistic forward simulations. However, this is a sizable task and efforts to add selection to projects like Standard PopSim are ongoing.

We agree that forward simulations would be a nice addition, but believe that it is a project in itself. Indeed, a major complication is that when, for computational tractability, purifying selection is simulated in small populations with realistic population-scaled parameters, the reduction in diversity due to selection at unlinked sites has a major effect on neutral diversity levels (see, e.g., Robertson 1961). We hope to address this issue in future work. Meanwhile, we note that the theory that we rely on has been tested against simulations in the past (e.g., Charlesworth et al., 1993; Hudson and Kaplan, 1995; Nordborg et al., 1996).

Author Response

Reviewer #1 (Public Review):

The relationship between genetic disease and adaptation is important for biomedical research as well as understanding human evolution. This topic has received considerable attention over the past several decades in human genetics research. The present manuscript provides a much more comprehensive and rigorous analysis of this topic. Specifically, the authors select a set of ~4000 human Mendelian disease genes and examine patterns of recent positive selection in these genes using the iHS and nSL tests (both haplotype test) for selection. They then compare the signals of sweeps to control genes. Importantly, they match the control set to the disease genes based upon many different genomic variables, such as recombination rate, amount of background selection, expression level, etc. The authors find that there is a deficit of selective sweeps in disease genes. They test several hypotheses for this deficit. They find that the deficit of sweeps is stronger in disease genes at low recombination rate and those that have more disease mutations. From this, the authors conclude that strongly deleterious mutations could be impeding selective sweeps.

Strengths

The manuscript includes a number of important strengths:

1) It tackles an important question in the field. The question of selection in disease genes has been very well-studied in the past, with conflicting viewpoints. The present study examines this topic in a rigorous way and finds a deficit of sweeps in disease genes.

2) The statistical analyses are rigorously done. The genome is a confusing place and there can often be many reasons why a certain set of genes could differ from another set of genes, unrelated to the variable of interest. Di et al. carefully match on these genomic confounders. Thus, they rigorously demonstrate that sweeps are depleted in disease genes relative to control genes. Further, the pipeline for ranking the genes and testing for significance is solid.

3) The Introduction of the manuscript nicely relates different evolutionary models and explanations to patterns that could be seen in the data. As such, the present manuscript isn't just merely an exploratory analysis of patterns of sweeps in disease genes. Rather, it tests specific evolutionary scenarios.

Weaknesses

1) The authors did not discuss or test a basic explanation for the deficit of sweeps in disease genes. Namely, certain types of genes, when mutated, give rise to strong Mendelian phenotypes. However, mutations in these genes do not result in variation that gives rise to a phenotype on which positive selection could occur. In other words, there are just different types of genes underlying disease and positive selection. I could think that such a pattern would be possible if humans are close to the fitness optimum and strong effect mutations (like those in Mendelian disease genes) result in moving further away from the fitness optimum. On the other hand, more weak effect mutations could be either weakly deleterious or beneficial and subject to positive selection. I'm not sure whether these patterns would necessarily be captured by the overall measures of constraint which the disease and non-disease genes were matched on.

We thank the reviewer for suggesting that alternative explanation. It is indeed important that we compare it with our own explanation. To rephrase the reviewer’s suggestion, it is possible that disease genes may just have a different distribution of fitness effects of new mutations. Specifically, mutations in disease genes might have such large effects that they will consistently overshoot the fitness optimum, and thus not get closer to this optimum. This would prevent them from being positively selected. Two predictions can be derived from this potential scenario. First, we can predict a sweep deficit at disease genes, which is what we report. Second, we can also predict that disease genes should exhibit a deficit of older adaptation, not just recent adaptation detected by sweep signals. Indeed, the decrease in adaptation due to (too) large effect mutations would be a generic, intrinsic feature of disease genes regardless of evolutionary time. This means that under this explanation, we expect a test of long-term adaptation such as the McDonald-Kreitman test to also show a deficit at disease genes.

This latter prediction differs from the prediction made by our favored explanation of interference between deleterious and advantageous variants. In this scenario, the sweep deficit at disease genes is caused by the presence of deleterious, and most importantly currently segregating disease variants. Because the presence of the segregating variants is transient during evolution, our explanation does not predict a deficit of long-term adaptation. We can therefore distinguish which explanation (the reviewer’s or ours) is the most likely based on the presence or absence of a long-term adaptation deficit at disease genes.

To test this, we now compare protein adaptation in disease and control genes with two versions of the MK test called ABC-MK and GRAPES (refs). ABC-MK estimates the overall rate of adaptation, and also the rates of weak and strong adaptation,and is based on Approximate Bayesian Computation. GRAPES is based on maximum likelihood. Both ABC-MK and GRPES have shown to provide robust estimates of the rate of protein adaptation thanks to evaluations with forward population simulations (refs). We find no difference in long-term adaptation between disease and control non-disease genes, as shown in new figure 4. This shows that the explanation put forward by the reviewer of an intrinsically different distribution of mutation effects at disease genes is less likely than an interference between currently segregating deleterious variants with recent, but not with older long-term adaptation. We even show in the new figure 4 that disease genes and their controls have more, not less strong long-term adaptation compared to the whole human genome baseline (new figure 4C). Also, disease genes in low recombination regions and with many disease variants have experienced more, not less strong long-term adaptation than their controls. Therefore, far from overshooting the fitness optimum due to stronger fitness effects of mutations, it looks like that these stronger fitness effects might in fact be more frequently positively selected in these disease genes.

We now provide these new results P15L418:<br /> “Disease genes do not experience constitutively less long-term adaptive mutations<br /> A deficit of strong recent adaptation (strong enough to affect iHS or 𝑛𝑆!) raises the question of what creates the sweep deficit at disease genes. As already discussed, purifying selection and other confounding factors are matched between disease genes and their controls, which excludes that these factors alone could possibly explain the sweep deficit. Purifying selection alone in particular cannot explain this result, since we find evidence that it is well matched between disease and control genes (Figures 2 and Figure 4-figure supplement 1). Furthermore, we find that the 1,000 genes in the genome with the highest density of conserved elements do not exhibit any sweep deficit (bootstrap test + block-randomized genomes FPR=0.18; Methods). Association with mendelian diseases, rather than a generally elevated level of selective constraint, is therefore what matters to observe a sweep deficit. What then might explain the sweep deficit at disease genes?

As mentioned in the introduction, it could be that mendelian disease genes experience constitutively less adaptive mutations. This could be the case for example because mendelian disease genes tend to be more pleiotropic (Otto, 2004), and/or because new mutations in mendelian are large effect mutations (Quintana-Murci, 2016) that tend to often overshoot the fitness optimum, and cannot be positively selected as a result. Regardless of the underlying processes, a constitutive tendency to experience less adaptive mutations predicts not only a deficit of recent adaptation, but also a deficit of more long-term adaptation during evolution. The iHS and nSL signals of recent adaptation we use to detect sweeps correspond to a time window of at most 50,000 years, since these statistics have very little statistical power to detect older adaptation (Sabeti et al., 2006). In contrast, approaches such as the McDonald-Kreitman test (MK test) (McDonald and Kreitman, 1991) capture the cumulative signals of adaptative events since humans and chimpanzee had a common ancestor, likely more than six million years ago. To test whether mendelian disease genes have also experienced less long-term adaptation, in addition to less recent adaptation, we use the MK tests ABC-MK (Uricchio et al., 2019) and GRAPES (Galtier, 2016) to compare the rate of protein adaptation (advantageous amino acid changes) in mendelian disease gene coding sequences, compared to confounding factors-matched non-disease controls (Methods). We find that overall, disease and control non-disease genes have experienced similar rates of protein adaptation during millions of years of human evolution, as shown by very similar estimated proportions of amino acid changes that were adaptive (Figure 5A,B,C,D,E). This result suggests that disease genes do not have constitutively less adaptive mutations. This implies that processes that are stable over evolutionary time such as pleiotropy, or a tendency to overshoot the fitness optimum, are unlikely to explain the sweep deficit at disease genes. If disease genes have not experienced less adaptive mutations during long-term evolution, then the process at work during more recent human evolution has to be transient, and has to has to have limited only recent adaptation. It is also noteworthy that both disease genes and their controls have experienced more coding adaptation than genes in the human genome overall (Figure 5A), especially more strong adaptation according to ABC-MK (Figure 5C). The fact that the baseline long-term coding adaptation is lower genome-wide, but similarly higher in disease and their control genes, also shows that the matched controls do play their intended role of accounting for confounding factors likely to affect adaptation. The fact that long-term protein adaptation is not lower at disease genes also excludes that purifying selection alone can explain the sweep deficit at disease genes, because purifying selection would then also have decreased long-term adaptation. A more transient evolutionary process is thus more likely to explain our results.”

Then P22L613: “More importantly, the fact that constitutively less adaptation at disease genes combined to more power to detect sweeps in low recombination regions does not explain our results, is made even clearer by the fact that disease genes in low recombination regions and with many disease variants have in fact experienced more, not less long-term adaptation according to an MK analysis using both ABC-MK and GRAPES (Figure 5F,G,H,I,J). ABC-MK in particular finds that there is a significant excess of long-term strong adaptation (Figure 4H, P<0.01) in disease genes with low recombination and with many disease variants, compared to controls, but similar amounts of weak adaptation (Figure 5G, P=0.16). It might be that disease genes with many disease variants are genes with more mutations with stronger effects that can generate stronger positive selection. The potentially higher supply of strongly advantageous variants at these disease genes makes it all the more notable that they have a very strong sweep deficit in recent evolutionary times. This further strengthens the evidence in favor of interference during recent human adaptation: the limiting factor does not seem to be the supply of strongly advantageous variants, but instead the ability of these variants to have generated sweeps recently by rising fast enough in frequency.”

2) While I think the authors did a superb job of controlling for genome differences between disease and non-disease genes, the analysis of separating regions by recombination rate and number of disease mutations does not seem as rigorous. Specifically, the authors tested for enrichment of sweeps in disease genes vs control and then stratified that comparison by recombination rate and/or number of disease mutations. While this nicely matches the disease genes to the control genes, it is not clear whether the high recombination rate genes differ in other important attributes from the low recombination rate genes. Thus, I worry whether there could be a confounder that makes it easier/harder to detect an enrichment/deficit of sweeps in regions of low/high recombination.

We thank the reviewer for emphasizing the need for more controls when comparing our results in low or high recombination regions. We have now compared the confounding factors between low recombination disease genes and high recombination disease genes, as classified in the manuscript. As shown in new supp table Figure 6 figure supplement 1, confounding factors do not differ substantially between low and high recombination disease genes, and are all within a range of +/- 25% of each other. It would take a larger difference for any confounding factor to explain the sharp sweep deficit difference observed between the low and high recombination disease genes. The only factor with a 35% difference between low and high recombination mendelian disease genes is McVicker’s B, but this is completely expected; B is expected to be lower in low recombination regions.

We now write P20L569: “Further note that only moderate differences in confounding factors between low and high recombination mendelian disease genes are unlikely to explain the sweep deficit difference (Figure 6-figure supplement 1).”

Regarding the potential confounding effect of statistical power to detect sweeps differing in low and high recombination regions, please see our earlier response to main point 2.

Reviewer #2 (Public Review):

This paper seeks to test the extent to which adaptation via selective sweeps has occurred at disease-associated genes vs genes that have not (yet) been associated with disease. While there is a debate regarding the rate at which selective sweeps have occurred in recent human history, it is clear that some genes have experienced very strong recent selective sweeps. Recent papers from this group have very nicely shown how important virus interacting proteins have been in recent human evolution, and other papers have demonstrated the few instances in which strong selection has occurred in recent human history to adapt to novel environments (e.g. migration to high altitude, skin pigmentation, and a few other hypothesized traits).

One challenge in reading the paper was that I did not realize the analysis was exclusively focused on Mendelian disease genes until much later (the first reference is not until the end of the introduction on pages 7-8 and then not at all again until the discussion, despite referring to "disease" many times in the abstract and throughout the paper). It would be preferred if the authors indicated that this study focused on Mendelian diseases (rather than a broader analysis that included complex or infectious diseases). This is important because there are many different types of diseases and disease genes. Infectious disease genes and complex disease genes may have quite different patterns (as the authors indicate at the end of the introduction).

We want to apologize profusely for this avoidable mistake. We have now made it clearer from the very start of the manuscript that we focus on mendelian non-infectious disease genes. We have modified the title and the abstract accordingly, specifying mendelian and non-infectious as required.

The abstract states "Understanding the relationship between disease and adaptation at the gene level in the human genome is severely hampered by the fact that we don't even know whether disease genes have experienced more, less, or as much adaptation as non-disease genes during recent human evolution." This seems to diminish a large body of work that has been done in this area. The authors acknowledge some of this literature in the introduction, but it would be worth toning down the abstract, which suggests there has been no work in this area. A review of this topic by Lluis Quintana-Murci1 was cited, but diminished many of the developments that have been made in the intersection of population genetics and human disease biology. Quintana-Murci says "Mendelian disorders are typically severe, compromising survival and reproduction, and are caused by highly penetrant, rare deleterious mutations. Mendelian disease genes should therefore fit the mutation-selection balance model, with an equilibrium between the rate of mutation and the rate of risk allele removal by purifying selection", and argues that positive selection signals should be rare among Mendelian disease genes. Several other examples come to mind. For example, comparing Mendelian disease genes, complex disease genes, and mouse essential genes was the major focus of a 2008 paper2, which pointed out that Mendelian disease genes exhibited much higher rates of purifying selection while complex disease genes exhibited a mixture of purifying and positive selection. This paper was cited, but only in regard to their findings of complex diseases. A similar analysis of McDonald-Kreitman tables3 was performed around Mendelian disease genes vs non-disease genes, and found "that disease genes have a higher mean probability of negative selection within candidate cis-regulatory regions as compared to non-disease genes, however this trend is only suggestive in EAs, the population where the majority of diseases have likely been characterized". Both of these studies focused on polymorphism and divergence data, which target older instances of selection than iHS and nSL statistics used in the present study (but should have substantial overlap since iHS is not sensitive to very recent selection like the SDS statistic). Regardless, the findings are largely consistent, and I believe warrant a more modest tone.

We thank the reviewer for their recommendation. We should have written more about what is currently well known or unknown about recent adaptation in disease genes, and in more nuanced terms. Instead of writing “Understanding the relationship between disease and adaptation at the gene level in the human genome is severely hampered by the fact that we don't even know whether disease genes have experienced more, less, or as much adaptation as non-disease genes during recent human evolution”, we now write in the new abstract:

“Despite our expanding knowledge of gene-disease associations, and despite the medical importance of disease genes, their recent evolution has not been thoroughly studied across diverse human populations. In particular, recent genomic adaptation at disease genes has not been characterized as well as long-term purifying selection and long-term adaptation. Understanding the relationship between disease and adaptation at the gene level in the human genome is hampered by the fact that we don’t know whether disease genes have experienced more, less, or as much adaptation as non-disease genes during the last ~50,000 years of recent human evolution.”

We also toned down the start of the introduction. We now write P3L74:

“Despite our expanding knowledge of mendelian disease gene associations, and despite the fact that multiple evolutionary processes might connect disease and genomic adaptation at the gene level, these connections are yet to be studied more thoroughly, especially in the case of recent genomic adaptation.”

Although we agree that others have made extensive efforts to characterize older adaptation or purifying selection at disease genes compared to non-disease genes, we still believe that our results are novel and more conclusive about recent positive selection. Our initial statement was however poorly phrased. To our knowledge, our study is the first to look at the issue using specifically sweep statistics that have been shown to be robust to background selection, while also controlling for confounding factors. These sweep statistics have sensitivity for selection events that occurred in the past 30,000 or at most 50,000 years of human evolution (Sabeti et al. 2006). This is a very different time scale compared to the millions of years of adaptation (since divergence between humans and chimpanzees) captured by MK approaches.

We also want to note that we did cite the Blekhman et al. paper for their result of stronger purifying selection in our initial manuscript. It is true however that we did not specify mendelian disease genes, which was confusing. We want to apologize again for it:

From the earlier manuscript: “Multiple recent studies comparing evolutionary patterns between human disease and non-disease genes have found that disease genes are more constrained and evolve more slowly (lower ratio of nonsynonymous to synonymous substitution rate, dN/dS, in disease genes) (Blekhman et al., 2008; Park et al., 2012; Spataro et al., 2017)”

“Among other confounding factors, it is particularly important to take into account evolutionary constraint, i.e the level of purifying selection experienced by different genes. A common intuition is that disease genes may exhibit less adaptation because they are more constrained (Blekhman et al., 2008)”

It is important to remember that, as we mention in the introduction, previous comparisons did not take potential confounding factors at all into account. It is therefore unclear whether their conclusions were specific to disease genes, or due to confounding factors. We have now made this point clearer in the introduction, as we believe that we have made a substantial effort to control for confounding factors, and that it is a substantial departure from previous efforts:

P7L201: “In contrast with previous studies, we systematically control for a large number of confounding factors when comparing recent adaptation in human mendelian disease and nondisease genes, including evolutionary constraint, mutation rate, recombination rate, the proportion of immune or virus-interacting genes, etc. (please refer to Methods for a full list of the confounding factors included).”.

P9L253: “These differences between disease and non-disease genes highlight the need to compare disease genes with control non-disease genes with similar levels of selective constraint. To do this and compare sweeps in mendelian disease genes and non-disease genes that are similar in ways other than being associated with mendelian disease (as described in the Results below, Less sweeps at mendelian disease genes), we use sets of control non-disease genes that are built by a bootstrap test to match the disease genes in terms of confounding factors (Methods)”.

Furthermore, we have now added a comparison of older adaptation in disease and non-disease genes using a recent version of the MK test called ABC-MK, that can take background selection and other biases such as segregating weakly advantageous variants into account. Also controlling for confounding factors, we find no difference in older adaptation between disease and non-disease genes (please see our response to main point 2).

Therefore, contrary to the reviewer’s claim that the sweep statistics and MK approaches should have substantial overlap, we now show that it is clearly not the case. We further show that the lack of overlap is expected under our explanation of our results based on interference between recessive deleterious and advantageous variants (see our responses to main point 1 and to reviewer 1 weakness 1).

Previous analyses were using much smaller mendelian disease gene datasets, less recent polymorphism datasets and, critically, did not control for confounding factors. We also note that reference 3 (Torgerson et al. Plos Genetics 2009) does not make any claim about recent positive selection in mendelian disease genes compared to other genes. Their dataset at the time also only included 666 mendelian disease genes, versus the ~4,000 currently known.

In short, we do think that we have a claim for novelty, but the reviewer is entirely right that we did a poor job of giving due credit to previous important work. These previous studies deserved much better credit than no credit at all. We want to thank the reviewer from avoiding us the embarrassment of not citing important work.

We now cite the papers referenced by the reviewer as appropriate in the introduction, based on the scope of their results:

P3L93: “Multiple recent studies comparing evolutionary patterns between human mendelian disease and non-disease genes have found that mendelian disease genes are more constrained and evolve more slowly (Blekhman et al., 2008; Quintana-Murci, 2016; Spataro et al., 2017; Torgerson et al., 2009). An older comparison by Smith and Eyre-Walker (Smith and Eyre-Walker, 2003) found that disease genes evolve faster than non-disease genes, but we note that the sample of disease genes used at the time was very limited.”

P5L134 “Among possible confounding factors, it is particularly important to take into account evolutionary constraint, i.e the level of purifying selection experienced by different genes. A common intuition is that mendelian disease genes may exhibit less adaptation because they are more constrained (Blekhman et al., 2008; Spataro et al., 2017; Torgerson et al., 2009),”

There are some aspects of the current study that I think are highly valuable. For example, the authors study most of the 1000 Genomes Project populations (though the text should be edited since the admixed and South Asian populations are not analyzed, so all 26 populations are not included, only the populations from Africa, East Asia, and Europe are analyzed; a total of 15 populations are included Figures 2-3). Comparing populations allows the authors to understand how signatures of selection might be shared vs population-specific. Unfortunately, the signals that the authors find regarding the depletion of positive selection at Mendelian disease genes is almost entirely restricted to African populations. The signal is not significant in East Asia or Europe (Figure 2 clearly shows this). It seems that the mean curve of the fold-enrichment as a function of rank threshold (Figure 3) trends downward in East Asian and European populations, but the sampling variance is so large that the bootstrap confidence intervals overlap 1). The paper should therefore revise the sentence "we find a strong depletion in sweep signals at disease genes, especially in Africa" to "only in Africa". This opens the question of why the authors find the particular pattern they find. The authors do point out that a majority of Mendelian disease genes are likely discovered in European populations, so is it that the genes' functions predate the Out-of-Africa split? They most certainly do. It is possible that the larger long-term effective population size of African populations resulted in stronger purifying selection at Mendelian disease genes compared to European and East Asian populations, where smaller effective population sizes due to the Out-of-Africa Bottleneck diminished the signal of most selective sweeps and hence there is little differentiation between categories of genes, "drift noise"). It is also surprising to note that the authors find selection signatures at all using iHS in African populations while a previous study using the same statistic could not differentiate signals of selection from neutral demographic simulations4.

We want to thank the reviewer profusely for putting us on the right track thanks to their insightful suggestion. As described in our response to reviewer 1 weakness 1, we have now shown with simulations that the interference of deleterious variants on advantageous variants is strongly decreased during a bottleneck of a magnitude similar to the Out of Africa bottlenecks experienced by East Asian and European populations. This decrease of interference is likely strong enough to not require any other explanation, even if other processes may also be at work, such as a decrease of the sweeps signals as suggested by the reviewer.

About the Granka et al. paper, the last author of the current manuscript has already shown in a previous paper (ref) that the type of approaches used to quantify recent adaptation is likely to be severely underpowered due to a number of confounding factors, notably including comparing genic and non-genic windows that are not sufficiently far from each other to not overlap the same sweep signals. Our result are also based on much more recent and less biased sets of SNPs used to measure the sweeps statistics.

The authors find that there is a remarkably (in my view) similar depletion across all but one MeSH disease classes. This suggests that "disease" is likely not the driving factor, but that Mendelian disease genes are a way of identifying where there are strongly selected deleterious variants recurrently arising and preventing positively selected variants. This is a fascinating hypothesis, and is corroborated by the finding that the depletion gets stronger in genes with more Mendelian disease variants. In this sense, the authors are using Mendelian disease genes as a proxy for identifying targets of strong purifying selection, and are therefore not actually studying Mendelian disease genes. The signal could be clearer if the test set is based on the factor that is actually driving the signal.

Based on the reviewer’s comment, we have now better explained why our results are unlikely to be a generic property of purifying selection alone. As we explain in our response to main point 3, our results cannot be explained by purifying selection alone, because we match purifying selection between disease genes and the controls. Indeed, we now show with additional MK analyses and GERP-based analyses that our controls for confounding factors already account for purifying selection. This is shown by the fact that disease genes and their controls have similar distributions of deleterious fitness effects.

In addition, we added a comparison that shows that purifying selection alone does not explain our results. Instead of comparing sweeps at disease and non-disease genes, we compared sweeps (in Africa) between the 1,000 genes with the highest density of conserved, constrained elements and other genes in the genome. If purifying selection is the factor that drives the sweep deficit at disease genes, then we should see a sweep deficit among the genes with the most conserved, constrained elements compared to other genes in the genome. However, we see no such sweep deficit at genes with a high density of conserved, selectively constrained elements (boostrap test + block randomization of genomes, FPR=0.18). See P15L424. Note that for this comparison we had to remove the matching of confounding factors corresponding to functional and purifying selection densities (new Methods P40L1131).

Again, our results are better explained not just by purifying selection alone, but more specifically by the presence of interfering, segregating deleterious variants. It is perfectly possible to have highly constrained parts of the genome without having many deleterious segregating variants at a given time in evolution.

The similarity across MeSH classes can be readily explained if what matters is interference with deleterious segregating variants. Because all types of diseases have deleterious segregating variants, then it is not surprising that different MeSH disease categories have a similar sweep deficit. We make that point clearer in the revised manuscript:

P26L707: “The sweep deficit is comparable across MeSH disease classes (Figure 8), suggesting that the evolutionary process at the origin of the sweep deficit is not diseasespecific. This is compatible with a non-disease specific explanation such as recessive deleterious variants interfering with adaptive variants, irrespective of the specific disease type.”.

One of the most important steps that the authors undertake is to control for possible confounding factors. The authors identify 22 possible confounding factors, and find that several confounding factors have different effects in Mendelian disease genes vs non-disease genes. The authors do a great job of implementing a block-bootstrap approach to control for each of these factors. The authors talk specifically about some of these (e.g. PPI), but not others that are just as strong (e.g. gene length). I am left wondering how interactions among other confounding factors could impact the findings of this paper. I was surprised to see a focus on disease variant number, but not a control for CDS length. As I understand it, gene length is defined as the entire genomic distance between the TSS and TES. Presumably genes with larger coding sequence have more potential for disease variants (though number of disease variants discovered is highly biased toward genes with high interest). CDS length would be helpful to correct for things that pS does not correct for, since pS is a rate (controlling for CDS length) and does not account for the coding footprint (hence pS is similar across gene categories).

Based on our response to the previous point, it is clear that a high density of coding sequences, or conserved constrained sequence in general are not enough to explain our results. Furthermore, we want to remind the reviewer that we already control for coding sequence length through controlling for coding density, since we use windows of constant sizes.

The authors point out that it is crucial to get the control set right. This group has spent a lot of time thinking about how to define a control set of genes in several previous papers. But it is not clear if complex disease genes and infectious disease genes are specifically excluded or not. Number of virus interactions was included as a confounding factor, so VIPs were presumably not excluded. It is clear that the control set includes genes not yet associated with Mendelian disease, but the focus is primarily on the distance away from known Mendelian disease genes.

We are sorry that we were not more explicit from the start of the manuscript. We now make it clearer what the set disease genes includes or not throughout the entire manuscript, by repeating that we focus specifically on mendelian, non-infectious disease genes. By noninfectious, we mean that we excluded genes with known infectious disease-associated variants. This does not exclude most virus-interacting genes since most of them are not associated at the genetic variant level with infectious diseases. It is also important to note that the effect of virus interactions is accounted for by matching the number of interacting viruses between mendelian disease genes and controls.

We write P29L818: “By non-infectious, we mean that we excluded genes with known infectious disease-associated variants. This does not exclude most VIPs since most of them are not associated at the genetic variant level with infectious diseases. It is important to note that the effect of virus interactions is accounted for by matching the number of interacting viruses between mendelian disease genes and controls.”

Minor comments:

On page 13, the authors say "This artifact is also very unlikely due to the fact that recombination rates are similar between disease and non-disease genes (Figure 1)." However, Figure 1 shows that "deCode recombination 50kb" is clearly higher in disease genes and comparable at 500kb. The increased recombination rate locally around disease genes seems to contradict the argument formulated in this paragraph.

We apologize for the lack of precision in this sentence. What we meant is that the recombination rates are not different enough that the mentioned hypothetical artifact would be able to explain our results. We also forgot to remind at this point in the manuscript that we match recombination between disease genes and controls. We now use more precise language:

P28L772 “The recombination rate at disease genes is also only slightly different from the recombination rate at non-disease genes (Figure 1), and we match the recombination rate between disease genes and controls.”.

Reviewer #3 (Public Review):

In this paper, the authors ask whether selective sweeps (as measured by the iHS and nSL statistics) are more or less likely to occur in or near genes associated with Mendelian diseases ("disease genes") than those that are not ("non-disease genes"). The main result put forward by the authors is that genes associated with Mendelian diseases are depleted for sweep signatures, as measured by the iHS and nSL statistics, relative to those which are not.

The evidence for this comes from an empirical randomization scheme to assess whether genes with signatures of a selective sweep are more likely to be Mendelian disease genes that not. The analysis relies on a somewhat complicated sliding threshold scheme that effectively acts to incorporate evidence from both genes with very large iHS/nSL values, as well as those with weaker signals, while upweighting the signal from those genes with the strongest iHS/nSL values. Although I think the anlaysis could be presented more clearly, it does seem like a better analysis than a simple outlier test, if for no other reason than that the sliding threshold scheme can be seen as a way of averaging over uncertainty in where one should set the threshold in an outlier test (along with some further averaging across the two different sweeps statistics, and the size of the window around disease associated genes that the sweep statistics are averaged over). That said, the particular approach to doing so is somewhat arbitrary, but it's not clear that there's a good way to avoid that.

In addition to reporting that extreme values of iHS/nSL are generally less likely at Mendelian disease genes, the authors also report that this depletion is strongest in genes from low recombination regions, or which have >5 specific variants associated with disease.

Drawing on this result, the authors read this evidence to imply that sweeps are generally impeded or slowed in the vicinity of genes associated with Mendelian diseases due to linkage to recessive deleterious variants, which hitchhike to high enough frequencies that the selection against homozygotes becomes an important form of interference. This phenomenon was theoretically characterized by Assaf et al 2015, who the authors point to for support. That such a phenomenon may be acting systematically to shape the process of adaptation is an interesting suggestions. It's a bit unclear to me why the authors specifically invoke recessive deleterious mutations as an explanation though. Presumably any form of interference could create the patterns they observe? This part of the paper is, as the authors acknowledge, speculative at this point.

We thank the reviewer for their comments. We are sorry that we did not provide a clear explanation of why only recessive deleterious mutations are expected to interfere more than other types of deleterious variants. This was shown by Assaf et al. (2015), and we should have stated it explicitly. The reason why recessive deleterious variants interfere more than additive or dominant ones is that they can hitchhike together with an adaptive variant to substantial frequencies before negative selection actually happens, when a significant number of homozygous individuals for the deleterious mutation start happening in the population. On the contrary dominant mutations do not make it to the same high frequencies linked to an adaptive variant, because they start being selected negatively as soon as they appear in the population.

We now write P18L496: “In diploid species including humans, recessive deleterious mutations specifically have been shown to have the ability to slow down, or even stop the frequency increase of advantageous mutations that they are linked with (Assaf et al., 2015). Dominant variants do not have the same interfering ability, because they do not increase in frequency in linkage with advantageous variants as much as recessive deleterious do, before the latter can be “seen” by purifying selection when enough homozygous individuals emerge in a population (Assaf et al., 2015).”

We have also confirmed with SLiM forward simulations that recessive deleterious variants interfere with adaptive variants much more than dominant ones (Table 1).

I'm also a bit concerned by the fact that the signal is only present in the African samples studied. The authors suggest that this is simply due to stronger drift in the history of European and Asian samples. This could be, but as a reader it's a bit frustrating to have to take this on faith.

We thank the reviewer for pointing out this issue with our manuscript. We have now shown, as detailed above in our response to main point 1, reviewer 1 weakness 1, that a weaker sweep deficit at disease genes in Europe and East Asia is an expected feature under the interference explanation, due to the weakened interference of recessive deleterious variants during bottlenecks of the magnitude observed in Europe and East Asia. We therefore believe that these new results strengthen our previous claim regarding the role interference between deleterious and advantageous variants. We want to thank the reviewer for forcing us to examine the difference between results in Africa and out of Africa, as the manuscript is now more consistent and our results substantially better explained.

There are other analyses that I don't find terribly convincing. For example, one of the anlayses shows that iHS signals are no less depleted at genes associated with >5 diseases than with 1 does little to convince me of anything. It's not particularly clear that # of associated disease for a given gene should predict the degree of pleiotropy experienced by a variant emerging in that gene with some kind of adaptive function. Failure to find any association here might just mean that this is not a particularly good measure of the relevant pleiotropy.

We agree with the reviewer that the number of associated disease may not be a good measure of pleiotropy. Unfortunately to our knowledge there is currently no good measure of gene pleiotropy in human genomes. Given that the evidence in favor of interference of deleterious variants is now strengthened, we have chosen to remove this analysis from the manuscript. As we now explain throughout the manuscript, pleiotropy is an unlikely explanation in the first place because of the fact that disease genes have not experienced less long-term adaptation (see the details on our new MK test results in the response to main point 2).

P16L447: “We find that overall, disease and control non-disease genes have experienced similar rates of protein adaptation during millions of years of human evolution, as shown by very similar estimated proportions of amino acid changes that were adaptive (Figure 5A,B,C,D,E). This result suggests that disease genes do not have constitutively less adaptive mutations. This implies that processes stable over evolutionary time such as pleiotropy, or a tendency to overshoot the fitness optimum, are unlikely to explain the sweep deficit at disease genes.”.

A last parting thought is that it's not clear to me that the authors have excluded the hypothesis that adaptive variants simply arise less often near genes associated with disease. The fact that the signal is strongest in regions of low recombination is meant to be evidence in favor of selective interference as the explanation, but it is also the regime in which sweeps should be easiest to detect, so it may be just that the analysis is best powered to detect a difference in sweep initiation, independent of possible interference dynamics, in that regime.

We thank the reviewer for stating these important alternative explanations that needed more attention in our manuscript. In our response to main point 2 above, we explain that higher statistical power in low recombination regions is unlikely to explain our results alone, because we also show that the sweep deficit is substantially present not only in low recombination regions, but also requires the presence of a higher number of disease variants. We also describe in our response to main point 2 how our new MK-test results on long-term adaptation make it very unlikely that mendelian disease genes experience constitutively less adaptation. We want to thank the reviewer again for pointing out this issue with our manuscript, since it was indeed an important missing piece.

Author Response

Reviewer 1

Panda and co-workers analyzed RS fMRI recordings from healthy patients and from two types of coma: UWS and MCS. They characterized the time-resolved functional connectivity in terms of metastability (time-variance of the Kuramoto order parameter), spatiotemporal patterns via non-negative tensor factorization, and its relationship to the eigenmodes of structural connectivity. Finding greater metastability and non-stationarity of the DMN network in healthy MCS patients, than in UWS patients, they found that the best discriminators to classify the different DoCs are the number of excursions (nonstability) from the DMN, salience and FPN networks extracted by the NNTF analysis. Interestingly, the data-driven NNTF yielded a novel sub-network comprising the FPN and some subcortical structures. The excursions and dwell times from this FPN subnetwork showed to be significantly lower in the UWS patients than in MCS. Surrogate data testing assures that the different methods and fits are effectively expressing the functional connectivity matrices measured.

Overall, I think that the results are correct and they advance in the characterization and understanding of the brain under DoC. However, some improvements can be made in the way the results, and the rationale behind them, are presented.

We thank Prof. Patricio Orio for his assessment.

While reading the Results section, it is easy to have the impression of a disconnected set of analyses that just happened to be together. In particular, the section about the structural eigenmodes and their relationship with the time-resolved FC seems to have little connection with the rest of the work, except for confirming (yet again) that DoC patients have a less dynamic FC. More elaboration about the relevance of these results, and what they say about DoC (that other dynamical FC analyses don't), is needed both in the introduction and discussion. Although a clear explanation is given in the introduction, the bottom line seems to be yet another measure of metastability. Perhaps, a better explanation of what underlies the 'modulation strength of eigenmodes expression' will be helpful for distinguishing this analysis from others. How novel is the connection that is being done with the structural connectivity and why is this important? Moreover, the eigenmodes analysis has little-to-none importance in the discrimination of patients done at the end; thus, its place within the big picture is hard to evaluate.

We understand the reviewer’s position. Part one of our work covers time-resolved FC and spatiotemporal networks in DoC. Part two covers the relationship between timeresolved FC and eigenmodes of the structural network. The rationale for including part two is the following: there is a lot of literature that shows that eigenmodes of the structural network can be considered as ‘building blocks’ or basis functions/vectors for spatiotemporal networks at the functional level (Aqil et al., 2021; Atasoy et al., 2016, 2018; Deslauriers-Gauthier et al., 2020; Gabay et al., 2018; Gabay and Robinson, 2017; Robinson et al., 2016; Robinson, 2021; Tewarie et al., 2019, 2020; Wang et al., 2017). Ideally to link part one and two, you would take this notion further by analysing if the magnitude eigenmode coefficients differed between UWS, MCS and healthy controls and how this would relate to dwell times or expression of spatiotemporal networks. However, this would lead to an immense multiple testing issue, which would be impossible to overcome with our sample size. An important link between part one and two of our work is the relationship between change in eigenmode expression and metastability. Our measure for metastability is only a proxy for metastability. Lack of change in eigenmode expressions seems to confirm this result of metastability.

To allow for better integration of part one and two of our work, we have added to the introduction:

“These eigenmodes can be considered as patterns of ‘hidden connectivity’ that come to expression at the level of functional networks. It has been postulated that eigenmodes form elementary building blocks for spatiotemporal dynamics (Aqil et al., 2021). There is evidence that the well-known resting state networks can be explained by activation of a small set of eigenmodes (Atasoy et al., 2018).”

We have also clarified in the result section:

“As resting-state network activity can be explained by activation of structural eigenmodes, we next analyse the role of fluctuations in eigenmode expression over time.”

Something that I find counter-intuitive and that may confuse some readers, is the (apparent) contradiction between the diminished metastability in the DoC conditions and the reduced dwell times (Figure S1; also "the inability to sequentially dwell for prolonged times in a different set of eigenmodes", as stated in the Discussion). Fewer excursions and shorter dwell times can only mean that some networks are just less visited and maybe this would be enough to distinguish between conditions. Further explaining this will help to understand better the implications of the work.

We understand the reviewer’s point, however we disagree that diminished metastability is in contradiction with the findings on dwell times. We show that dwell times are reduced in the posterior DMN, FPN and sub-FPTN networks, however, there is very long dwelling in the residual network in DoC. Hence, the brain resides in fewer network states in DoC, which is in agreement with reduced metastability. Our proxy for metastability is the standard deviation of the Kuramoto order parameter. Whenever there are more visits to network states, or switching between network states as is the case for healthy controls in our data, this would lead to phase uncoupling followed by phase synchronization, which would hence boost the standard deviation of the Kuramoto order parameter (a proxy for metastability).

We agree with the reviewer that the sentence starting “the inability to sequentially dwell for prolonged….” Is confusing. We have now removed this statement.

We have now added to the result section:

“These findings of very short dwell times in the posterior DMN, FPN and sub-FPTN and long dwell time in the residual network can be considered as a contraction of the functional network repertoire in DoC, which is in agreement with a loss in metastability in these patients.”

Finally, some comments about the connection(s) of these analyses with the commonly used FCD analysis (based on sliding windows of pair-wise correlations) will be useful, to put better this work into the big picture of time evolution of the functional connectivity.

We have now discussed sliding window-based analysis in the context of our work in the methodology section.

“Lastly, we have used a high temporal resolution method to estimate time-resolved connectivity at every time point instead of a sliding window-based method. Previous studies using sliding window approaches have provided novel insights into brain dynamics of loss of consciousness, such as the brain co-occurrence of functional connectivity patterns, which is known as brain states and its temporal (i.e., rate of pattern occurrence (probability) and between pattern transition probabilities) alteration in loss of consciousness in DoC patients (Demertzi et al., 2019) and anaesthesia induced loss of consciousness (Barttfeld et al., 2014a; Uhrig et al., 2018). However, sliding window approaches have limited sensitivity to non-stationarity in the fMRI BOLD signals (Hindriks et al., 2016) and lack to provide spatial alteration of classical brain functional network. The exploration of the spatiotemporal aspects of well-known resting state networks is an important step forwards for better understanding the relation between brain function and consciousness, in a way that is impossible to achieve at the whole brain level. In addition, recent work on time-resolved connectivity shows that brief periods of co-modulation in BOLD signals are an important driving factor for functional connectivity (Esfahlani et al., 2020; Hindriks et al., 2016).”

Reviewer 2

The study is of high significance, rigor, and novelty. Despite the many studies of repertoire, dynamic connectivity, etc., in the study of consciousness, there is (surprisingly, as I confirmed with a literature search) a dearth of application of these approaches to disorders of consciousness. The manuscript is well-written and transparent about its limitations. The author should consider the following recommendations:

We thank the reviewer for his/her assessment of our work.

1) There is frequent reference to "subcortical" and related networks, but I see no description in the text of which subcortical structures are involved. Panel N of figure 2 is helpful but I think that more explicit detail is important, especially given the specific predictions of mesocircuit theory.

We have provided details for the subcortical networks presented in the Panel N of Figure 2. In the manuscript we provide a textual description of the brain areas that are part of the network. To improve the clarity of the description of the network, we also now refer to it as “subcortical fronto-temporoparietal (Sub-FTPN)”.

In the result section, it read as: “This modulated subcortical fronto-temporoparietal network consist of the following brain regions: bilateral thalamus, caudate, right putamen, bilateral anterior and middle cingulate, inferior and middle frontal areas, supplementary motor cortex, middle and inferior temporal gyrus, right superior temporal, bilateral inferior parietal and supramarginal gyrus.”

2) Similarly, although the global neuronal workspace does posit a critical role for recurrent frontal-parietal networks, can the authors be more specific about the nodes of the proposed workspace and what they found empirically?

As above mentioned, we have provided more details about the regions part of the “subcortical fronto-temporoparietal”. As the reviewers rightfully noted, this network also shows some overlap with the Global Neuronal Workspace. We refer to that in more detail in the discussion, highlighting how our functional networks overlap and differ with the two networks (i.e., one feedforward only, one with recurrent activity), and with the predictions of the mesocircuit model. For more detail, please refer to the reply to point 1 of “Recommendations for the authors”.

3) The classification sensitivity/specificity did not, in my opinion, add much to the manuscript, especially since the number of patients is not remotely close to what would be required for a population-based diagnostic approach. If the authors chose to include this with any reference to diagnosis (highlighted in the introduction and elsewhere), I would encourage a comparison with similar data from other clinical or neuroimagingbased diagnostic approaches. However, I think the value of the study resides more with mechanistic understanding than diagnosis.

We agree with your suggestions that the primary aim of our work is to provide a mechanistic understanding of loss of consciousness. Therefore, we have removed the classification part from the paper and explain our findings focusing on mechanism of pathological unconsciousness rather than its potential as a clinical diagnostic tool. This change has required several textual edits throughout the manuscript.

Review #1 Public Review

Panda and co-workers analyzed RS fMRI recordings from healthy patients and from two types of comma: UWS and MCS. They characterized the time-resolved functional connectivity in terms of metastability (time-variance of the Kuramoto order parameter), spatiotemporal patterns via non-negative tensor factorization, and its relationship to the eigenmodes of structural connectivity. Finding greater metastability and non-stationarity of the DMN network in healthy MCS patients, than in UWS patients, they found that the best discriminators to classify the different DOCs are the number of excursions (non-stability) from the DMN, salience and FPN networks extracted by the NNTF analysis. Interestingly, the data-driven NNTF yielded a novel sub-network comprising the FPN and some subcortical structures. The excursions and dwell times from this FPN sub-network showed to be significantly lower in the UWS patients than in MCS. Surrogate data testing assures that the different methods and fits are effectively expressing the functional connectivity matrices measured.

Overall, I think that the results are correct and they advance in the characterization and understanding of the brain under DOC. However, some improvements can be made in the way the results, and the rationale behind them, are presented.

While reading the Results section, it is easy to have the impression of a disconnected set of analyses that just happened to be together. In particular, the section about the structural eigenmodes and their relationship with the time-resolved FC seems to have little connection with the rest of the work, except for confirming (yet again) that DOC patients have a less dynamic FC. More elaboration about the relevance of these results, and what they say about DOC (that other dynamical FC analyses don't), is needed both in the introduction and discussion. Although a clear explanation is given in the introduction, the bottom line seems to be yet another measure of metastability. Perhaps, a better explanation of what underlies the 'modulation strength of eigenmodes expression' will be helpful for distinguishing this analysis from others. How novel is the connection that is being done with the structural connectivity and why is this important? Moreover, the eigenmodes analysis has little-to-none importance in the discrimination of patients done at the end; thus, its place within the big picture is hard to evaluate.

Something that I find counter-intuitive and that may confuse some readers, is the (apparent) contradiction between the diminished metastability in the DOC conditions and the reduced dwell times (Figure S1; also "the inability to sequentially dwell for prolonged times in a different set of eigenmodes", as stated in the Discussion). Fewer excursions and shorter dwell times can only mean that some networks are just less visited and maybe this would be enough to distinguish between conditions. Further explaining this will help to understand better the implications of the work.

Finally, some comments about the connection(s) of these analyses with the commonly used FCD analysis (based on sliding windows of pair-wise correlations) will be useful, to put better this work into the big picture of time evolution of the functional connectivity.

3 OS recommendations for old laptops: 1. Puppy Linux 2. AntiX 3. Lubuntu

Reviewer #2 (Public Review):

Zhou et al. investigates whether alpha-band (8-14 Hz) neural oscillations differentially modulate sensory signal and noise during visual detection. The authors reason that the preferential modulation of signal predicts a relationship between alpha power and the subject's perceptual discriminability but not decision criterion, and conversely, that the similar modulation of signal and noise predicts a relationship between alpha power and the subject's decision criterion but not perceptual discriminability. The authors find that alpha power in early visual cortex does not correlate with the block-wise changes in the subjects' decision criterion. However, trial-to-trial variations in visual cortical alpha power during the trial period before stimulus presentation correlate inversely with the subject's perceptual discriminability. Moreover, lower prestimulus alpha power in visual areas is associated with enhanced information that can be decoded about the visual stimulus from recorded neural activity. Finally, the subject's accuracy depends on the phase of the alpha oscillations in parietal and frontal regions. Based on these findings, the authors conclude that alpha power modulates sensory signals more strongly than noise.

The question is interesting, the task design and priming procedures are rigorous and clever, and the analyses are sophisticated.

The conclusions of the paper would be more strongly supported if the concerns below could be addressed:

1. A potential strength of the manuscript consists in correlating alpha band power with not only the subject's accuracy (i.e. percentage of trials correct) but with the indices of d' and decision criterion from signal detection theory. Because any difference in accuracy can depend on a difference in only d', only criterion, or both, using d' and criterion provide a more precise quantification of behavior. However, this strength is undermined by the unquantified statistical bias in the estimates of d' and criterion as a result of limited sample size of certain categories of trials. This is an issue to which the authors themselves allude (at the bottom of page 4 and top of page 5) and is well documented (Macmillan and Creelman, 2004), but it is not addressed quantitatively in the manuscript. This issue therefore raises concern about measurements of d' and criterion throughout the manuscript. Because the manuscript's conclusions depend critically on the measurements of the subject's d' and criterion, therefore concern is also raised about the manuscript's conclusions.

2. Another potential strength of the manuscript involves shifting of the subject's decision criterion between blocks without changes in the subject's d' to isolate the relationship between decision criterion and alpha oscillations. However, the analyses in Figure 2C indicate that the subject's d' changed between blocks, and the brief argument that this difference in d' should be ignored is not quite convincing without detailed quantitative analyses. Because the possibility remains that d' changed between priming conditions, and yet no difference in alpha power could be detected between conditions, the result appears to be inconsistent with the finding that lower alpha power is related to enhanced d'.

3. The interpretations of the data at four places in the manuscript seem to be overly focused on a limited set of conditions or time windows, while not taking into account other conditions or time windows. This makes the interpretations appear incomplete and weakens confidence in the conclusions. (A) If trial-to-trial variations in alpha power are inversely correlated with d', then one should expect to see this relationship not only in the conservative priming condition but also in the liberal priming condition (Figure 3D). The absence of a significant relationship in the liberal priming condition appears to be inconsistent with the conclusions of the paper and is not addressed. (B) While the trial-to-trial prestimulus alpha power is explored for its relationship with either d' or criterion, the trial-to-trial alpha power during other task periods, in particular during stimulus presentation or during mask presentation, is not examined for its relationship with d' or criterion. (C) A significant relationship between the subject's accuracy and the phase of the alpha oscillations in visual ROIs is detectable at multiple brief time points before stimulus onset (Figure 5A), yet the authors state in the discussion that alpha phase in visual areas does not modulate the subject's accuracy. Instead, the authors focus on the relationship between the subject's accuracy and the alpha phase in parietal and frontal areas. (D) The prestimulus alpha power in visual cortex is significantly related to the criterion of the MVPA classifier during the conservative priming condition (page 7, "beta_c = 0.0096, p = 0.016"). This appears inconsistent with the conclusion that alpha power is independent of decision criterion.

• THANK YOU, your idea
• ME:
• look for HelpPane: c:\windows; its a Link
• BUT:
• Properties/Security:
• 1- change owner=Adminsitrators
• 2- permisssions: Adminis=Full control
• 3- rename file: *.[NO]
• Test it!
• Never more!!!

Like NEXT OS?

• OK

• DISCUTIBLE!!!
• LET USERS DECIDE!!!

• ALL RIGHT
• TIRED OF "THESE" CHANGES

I wonder what I questioned that he couldn't defend.

Pretty impressive for a "Senior Editor" to be so incapable of defending his positions.

• from : https://hyp.is/11avpgIXEe2-1eNzya0k-A/bafkreid7fzg5f6jb7rugtsu2lu33hzkvs4xfaoxzr5b75swqzandabzije.ipfs.dweb.link/?filename=Textile%20Threads.pdf

Reviewer #1 (Public Review):

Here the authors examined volitional neural signals during an un-cued lever pulling task. The authors impressively monitored cortical activity using widefield Ca++ imaging over many sessions (days to months).

Mice received water-rewards to motivate them to pull the lever. The major aim in this study was to understand when neural signals corresponding to the upcoming level pull appeared within the cortex. This is an important question in sensorimotor control, namely, how and where do volitional signals associated with our actions arise in the brain? The authors compare their results at various points to human motor control, where readiness potentials appear prior to the execution of movement. Thus, the authors' study could make a meaningful contribution to understanding how neural activity changes prior to the initiation of a voluntary movement (i.e. there is no cue in their task).

Prior to each lever pull, neural activity exhibited oscillatory patterns that sharpened in amplitude with proximity to the pulling event. These oscillations could be observed throughout the cortex: in retrosplenial, barrel, somatosensory, visual, and motor regions. As previously reported, neural activity exhibited a reduction in variance prior to movement initiation. The collapsing in variance was observed prior to the movement in all areas, excepting visual cortex. These changes in variance were echoed by convex hull analyses, which aimed to summarize the space spanned by pre-pull neural activity. As the movement approached, the convex hull gradually narrowed. The intersection between the lever pull's convex hull and the convex hull associated with all paw movements appeared to decrease with training in the task. This suggested a restructuring in neural activity whereby lever pull movements became more distinct in their neural activity patterns.

To understand whether pre-pull neural activity was associated with the upcoming movement, the authors used an SVM to predict whether neural activity over some pre-pull window could predict the upcoming lever pull. They observed that SVM classifiers could indeed predict the upcoming action well in advance of the behavior, in some cases 10-15 sec prior to the lever pull. This result is quite notable, given that previous evidence in humans suggests that readiness potentials arise 0.5-1.5 seconds prior to movement. Thus, the authors' study suggests a much longer time horizon for volitional signals in the brain.

The authors' question is both intriguing and important to the field of motor control, but certain details about their task complicate interpretations of their data. Most importantly, in the pre-pull period, behavior was not generally quiescent. Because the task did not use a cue, animals engaged in many behaviors in the windows preceding rewarded lever pulls. Thus, it is hard to know whether pre-pull neural activity relates to the upcoming rewarded lever pull, or earlier lever pull events (and other behaviors) that were likely to occur within the SVM window itself. While the authors used a 3-second lockout in their analysis (only considered rewarded pulls that were not preceded by lever pulls in the past 3 seconds), it remains challenging to interpret neural activity prior to threshold value (and it is currently unclear whether this lockout period excludes all lever pulls, or only some that met certain criteria). Along these lines, when the lockout window was extended in a control analysis, the SVM's time horizon for volitional signals shortened, suggesting that pre-pull behaviors indeed influenced the primary results. Thus, it remains unclear exactly when volitional signals arise in this task. The authors could greatly strengthen their paper with additional neural and behavioral analyses on these matters.

• In W10 Pro (20H2)
• Error if 32bits
• DLLs 64b required
• Install 64bits version

• ANOTHER SOLUTION
• rename target in the link properties

7

8.1

DOI: 10.7554/eLife.69013

Resource: XPP-Aut: X-Windows Phase Plane plus Auto (RRID:SCR_001996)

Curator: @scibot

SciCrunch record: RRID:SCR_001996

What is this?

Kerzin differentiates between existence and intrinsic existence. Intrinsic existence is what the Buddha and what Nagarjuna is trying to negate.

Rovelli makes a good point about a prevalent attitude that science offers a truer perspective than common sense, while Nagarjuna is pointing out that even the scientific explanation is not the final one. For one thing, it implicitly depends on the existence of a reified self who is the ultimate solidified existing agent and final authority, which Nagarjuna negates with his tetralemma.

instalar también si tienes windows y también install.packages("av")

修改 Docker 运行中 Container 的映射端口。

（1）停止服务

停止容器服务

docker stop <container id>

停止 docker 服务 (Linux)

systemctl stop docker

（2）修改配置

查看 container 的 id hash 值

docker inspect <container_name>

C:\Users\xxj87>docker inspect b61792d860f2 [ { "Id": "b61792d860f24c7ba47f4e270e211736a1a88546375e97380884c577d31dab66", "Created": "2022-07-01T07:46:03.516440885Z", "Path": "/bin/sh",

配置目录

[nux]: cd /var/lib/docker/containers/4fd7/

修改文件 hostconfig.json 中的 PortBindings

vim hostconfig.json

"PortBindings":{"2222/tcp":[{"HostIp":"","HostPort":"2222"}],"5000/tcp":[{"HostIp":"","HostPort":"5000"}],"80/tcp":[{"HostIp":"","HostPort":"40001"}],"8070/tcp":[{"HostIp":"","HostPort":"8070"}],"8081/tcp":[{"HostIp":"","HostPort":"8081"}]},

"80/tcp":[{"HostIp":"","HostPort":"40001"}] 80 容器内部端口 40001 外部映射端口

修改 config.v2.json 中的 ExposedPorts

vi config.v2.json "ExposedPorts":{"2222/tcp":{},"5000/tcp":{},"80/tcp":{},"8081/tcp":{},"8070/tcp":{}},

重启服务

systemctl start docker

启动容器

docker start <container id>

验证修改

docker ps -a

There is no inherent virtue in insisting that paths continue to use backslash even though we're well past the days of CP/M and DOS. There are good reasons not to use it, however.

I do recognize that it provides a source of the type of obscurantism that Windows users take delight in. At this point, though, backslash-as-path-separator is a liability, and application authors should work to eradicate it from every UI surface that users come in contact with. Microsoft themselves should scrub their own apps so that e.g. even Windows Explorer favors the common solidus in any displayed path name.

Consider this to be a straw proposal for an "are we slash yet?" movement.

Test annotate

Transformative change is required across all aspects of society. With such short time windows, leverage points become critical.

• ASUS: McAfee pre-installed

looked off behind, into splendid openness and the range of the afternoon light: The West-facing design of Mr Osmond’s villa is anachronistic in its 19th-century context. Though some Roman structures opted for West-facing windows that could be opened and regulate temperature in cooler months, during the time James wrote the morning sun was considered preferable[1]. It was far less intense and helped one get out of bed. Indeed, there is copious scientific evidence demonstrating the importance of directly exposing one’s eye to sunlight in the early morning as a way to wake up and begin work at sunrise[2]. Thus, West-facing houses symbolise the past in several ways. They look backwards, watching the sun sink below the horizon, and the end of a day or period. They also represent an out-of-date architectural style and logistical emphasis on keeping a house warm. Finally, they are constructed for the purpose of leisure. Though offering excellent views late in the day, this would overheat most after work and come at the cost of being woken up to attend to business in the morning. James therefore parallels the preferences of this house’s tenant – Osmond’s indolence and pre-occupation with antiquity and tradition. [1] Verticchio, Elena, et al. "Conservation risks for paper collections induced by the microclimate in the repository of the Alessandrina Library in Rome (Italy)." Heritage Science 10.1 (2022): Pp. 1-15. [2] Ode, Koji L., and Hiroki R. Ueda. "The Flow of Time Inside the Cell: The Time of Days Given by Molecules Driving the Circadian Clocks." Minorities and Small Numbers from Molecules to Organisms in Biology. Springer, Singapore, 2018. 135-143

情景：单独下载了 Dart SDK

执行： where/where.exe flutter dart

出现提示 Flutter SDK 内的 dart 命令不在首位。

则需要更新 PATH，将 C:\path-to-flutter-sdk\bin 放在 C:\path-to-dart-sdk\bin\ 前面（当前场景）。 重启命令行使修改生效，再次运行 where，此时来自相同目录的 flutter 和 dart 已经排在前面。

windows映射docker

asdfasdf asd fasdf

Indeed, Wright is right to ask: Is our modern human civilization the greatest progress trap of all?

Exponential technological progress has shortened the time for dangerous levels of resource extraction and pollution loads to the extent that we face the potential of cascading global tipping points and enter a "hothouse earth" state: https://www.pnas.org/doi/10.1073/pnas.1810141115

Were this to happen, there is no place on earth that would be immune.

In hindsight, the unfortunate but predictable trend is one of every increasing size of progress traps, and ever shorter time windows when serious impacts occur. Today, it appears we have reached the largest size progress trap possible on a finite planet.

curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.1/install.sh | bash

sudo apt-get install curl

running multiple version of node

• skip :

• for : https://knovigator.com/quest/setup-node-on-wsl-3gv36qnm

install wsl from powershell

So in Google Docs this now works (in June 2022): Alt-i r

This is very interesting, as I had studied these theories in a few of my criminal justice courses. These theories link certain characteristics of neighborhoods to crime levels.

Author Response

Reviewer #1 (Public Review):

LaRue, Linder and colleagues present an automation (GLO-Bot) and analysis pipeline building on the previously developed GLO-Roots, which makes use of a constitutively expressed luciferase gene to image plant roots in thin soil containers (rhizotrons). After validation of the system using a set of 6 accessions, the authors then take advantage of the increased throughput to phenotype root system architecture (RSA) of 93 natural Arabidopsis accessions and perform genome-wide association to identify polymorphic genomic regions that are associated with specific RSA traits. I appreciate that the authors made all data available via zenodo.

The authors succeeded in automating the GLO-Root system. Overall, the GLO-Bot appears to be a nice platform to collect time-lapse images of root growth in soil-substrate using rhizotrons. The automation of the GLO-Roots system using the GLO-Bot is well described, although not in sufficient detail to be rebuilt by interested researchers, e.g. the software controlling the robot is not described or made available, precluding wide adoption of the method. The image processing pipeline is clearly described in the methods and in Figure 2. The pipeline open source and available for use and appears to work well overall, although in some cases the vector representation of the root system appears to be incomplete.

We thank reviewer #1 for raising these concerns. We have now made the general code for the software available (GitHub: https://github.com/rhizolab/rhizo-server). In addition, we uploaded the rhizotron laser cutting files (Zenodo DOI: https://doi.org/10.5281/zenodo.6694558) that would facilitate rebuilding the robot.

We understand the concerns about the vector representations of the root system.

These root system structures visible on the GLO-Bot images are indeed disconnected in many locations, due to variability in the reporter’s intensity and obstruction of the light path by soil particles. For traits like root angle, the disconnected nature of the root system is much less impactful as this method naturally uses “segments” of the root as individual elements for angle measurements.

The authors then present a quantitative analysis of RSA using a set of 93 accessions, with 6 replicates per accession, generating a large dataset on the diversity of RSA in Arabidopsis. Using average angle per day, the authors identify SNPs that significantly associated with angle at 28 days after sowing, and they describe a correlation between this trait and the mean diurnal temperature range at the site where the accession was originally collected. The main weakness of the manuscript in its current form are some details of the quantitative genetic analysis. In my opinion the quantitative genetic analysis would benefit from additional quality control as there are peculiarities in the dataset that was used as the basis for GWAS.

We understand the concerns from reviewer #1 about the quantitative genetic analysis. Ultimately, we performed the analyses in the way we explained in the paper with careful consideration. We have added in additional descriptions of the rationale for chosing certain methods that hopefully elucidate why we did the analyses in the way we did. We hope this paper serves as a resource for others to pursue additional studies on traits relevant to their research.

Reviewer #2 (Public Review):

Therese LaRue and colleagues have developed a second generation of the GLO-Roots system that had been developed in their lab and published in 2015. Importantly, the new system (GLO-Bot) and the analysis of the resulting images has now been largely automated and therefore provides a throughput allowing for genetic studies. In an impressive endeavor the authors have transformed more than 100 diverse accessions that had been selected using sensible criteria with the luciferase construct, which then allowed the RSA of these accessions to be measured using the GLO-Bot system. On a set of 6 diverse accessions, the authors carefully identify meaningful RSA traits that they then quantified in the accessions of a larger panel of almost 100 accessions. They also benchmarked the new imaging processing tools against gold-standard manual tools. Overall, they show that the data acquisition and analysis is reproducible and reasonably accurate. They then proceeded to conduct GWAS using the RSA traits and identified several significantly associated candidate SNPs. Finally, they correlated the RSA with environmental variables and found interesting correlations that are consistent with prior studies.

Strengths:

The manuscript presents interesting root phenotyping technology, a comprehensive atlas of RSA under rhizotron lab conditions in Arabidopsis, candidate genes potentially underlying RSA traits, and interesting associations of RSA and climate variables. This will be inspiring and useful to many other researchers and has the potential to be explored further in future studies.

We thank the reviewer for the encouraging feedback.

Weaknesses:

Some aspects of the data analyses are not well described and should be described more. The trait data is heavily processed to "breeding values" and it is a bit unclear when unprocessed and processed trait data is used and why. Also, limitations and caveats are not discussed sufficiently. For instance, presenting and discussing the issues and caveats of measuring RSA that was generated in thin and not very wide soil sheets using the GLO-Bot system when natural growth in soil is usually largely unconstrained. Moreover, the analysis of potential candidate genes from the GWAS is not very well developed. Finally, the trait data was not available with the manuscript and a major impact of a resource like this will come from the data being fully available to the community.

We appreciate the broad comments on the manuscript and have tried to address them through the specific responses below. Overall we believe the approaches we used are effective but with specific caveats and have used the revision as a means of better communicating the limitations of the approaches chosen.

Reviewer #3 (Public Review):

The authors provide a thorough description of a method to transform plants to be bioluminescent upon applications of the require substrate such that roots are visible on the windows of rhizoboxes. They have expanded on previous work by automatic the imaging process with a robot that moves rhizoboxes to an imager where images are captured. They have improved the image analysis pipeline to be mostly automated with a user presumably needed to run various scripts in batch mode on directories of images. One novel aspect of the image analysis pipeline is in using image subtraction to subtract the previous time root system from the current in order to identify new growth.

We thank the reviewer for highlighting the strengths of the manuscript.

Overall, I think the authors provide a great amount of detail in parts needed and the methods, but some recommendations to increase reproducibility are more information about actual root traits measured. For example, one concern would be if root length is only summing pixels without considering diagonal pixels having a length of square-root of two, sqrt(2).

This is a valid concern, rather than just summing the pixels, the length of the segments is actually calculated using the “Feret Diameter” (or caliper length) function in imageJ which does take diagonals into consideration

While the methodological aspects of the paper are compelling, the authors have furthered the significance through a biological application for genetic analysis among accessions of Arabidopsis and correlating root traits to climatic 'envirotypes' or data from the origin site of the respective accession. This genetic analysis would be furthered by greater consideration of time series analysis and multi-trait analysis, which is possible in GEMMA. The authors could consider genetic analysis of the PCA traits as well. Given the novelty of this type of time-series, multi-trait data - the authors can reach further here.

Absolutely, PCA approaches to disentangle the phenotype space would be highly interesting to further investigate, which we started in the Supplemental Figure 8. This figure decomposes all the data points including replicates and temporal values of the same replicate. The PC1 therefore mostly captures how plants change over time, while PC2 seems to capture the main trade-off of wide/horizontal vs deep/vertical root architectures that we describe throughout the text. We could make use of this PC space to quantify the average value per genotype in PC2 and utilize this value for GWA, although it is not obvious how replicated and temporal measurements behave in PCA and what would be its consequences when computing a genotype value. There will definitely be interesting work that we aim to pursue in this direction in the future.

Regarding the additional capabilities of GEMMA. We are not aware of a subtool that is able to analyze time series directly in GEMMA, but we will look into it. The multi-trait analysis in GEMMA is also interesting. We have utilized the multi-trait feature in the past, but this is limited to very few traits. We have 8 time points, thus 8 traits. For reference, when we have run multi-trait LMM with 2 traits, we have typically seen runtimes of ~9 days in large clusters. New tools continue to emerge in the field of quantitative genetics, such as the use of summary statistics of multiple GWAs to gain new insights, which we will pursue in the future. We have added possible future directions to the discussion section (page 14).

As far as the general structure of the manuscript, I struggled with the results mixing in the methods such that I was never sure if the lack of detail in methods there would be addressed later, along with the mixture of discussions. Perhaps these are personal choices, but the methods were also after supplemental. I simply ask the authors to consider the reader here by being honest with my own experience reading this manuscript.

We appreciate this comment of reviewer #3. Since this is a “Tools and Resources” article, we believe that a substantial part of the results section should include the methods that were applied. The methodology mentioned in the results section should always help the reader to understand the illustrated results in the figures. If readers would like to apply certain methods, however, more details can be found in the materials and methods section. We apologize if this was not always successful and led to confusion. In the final formatted version, all supplemental figures would be linked to the main figures so that the materials and methods section would follow the discussion.

Overall, I believe this manuscript advanced root phenotyping by providing relatively high-throughput (imaging is slow due to the long exposure times) data and doing the time-series, multi-trait genetic mapping. The authors mention imaging shoots but no data is presented - presumably, it would be interesting to tie that in but they may be reasons to not. The authors could also discuss more the advantages of this approach relative to color imaging that has also advanced significantly since the original GLO-Root paper was released. Last, I am not sure the description of the 6 accessions study adds much value to the paper, and probably many other preliminary studies were done to prototype. Overall, this is fantastic and substantial work presented in a compelling way.

Unfortunately, the shoot images that were taken did not have sufficient quality for further analysis and due to technical problems, the set of shoot images is not complete. We removed the part of shoot imaging from the text. It now reads:”Inside the imaging system, the rhizotrons were rotated using a Lambda 10-3 Optical Filter Changer (Sutter Instrument®, Novato, CA). If it was the first imaging day or a designated luciferin day (every six days), GLO-Bot added 50 mL of 300 μM D-luciferin (Biosynth International Inc., Itasca, IL) to the top of each rhizotron immediately before loading the rhizotron into the imager.”

The advantages of the GLO-Roots method over color imaging is clearly that the GLO-Roots method can capture a more complete image of root systems with finer roots (like Arabidopsis). We have added the possibility of using RGB imaging for bigger root systems to the discussion section (page 13).

pathping

pathping /n baidu.com

Reviewer #2 (Public Review):

Schubert et al. describe a new pooled screening strategy that combines protein abundance measurements of 11 proteins determined via FACS with genome-wide mutagenesis of stop codons and missense mutations (achieved via a base editor) in yeast. The method allows to identify genetic perturbations that affect steady state protein levels (vs transcript abundance), and in this way define regulators of protein abundance. The authors find that perturbation of essential genes more often alters protein abundance than of nonessential genes and proteins with core cellular functions more often decrease in abundance in response to genetic perturbations than stress proteins. Genes whose knockouts affected the level of several of the 11 proteins were enriched in protein biosynthetic processes while genes whose knockouts affected specific proteins were enriched for functions in transcriptional regulation. The authors also leverage the dataset to confirm known and identify new regulatory relationships, such as a link between the SDS amino acid sensor and the stress response gene Yhb1 or between Ras/PKA signalling and GAPDH isoenzymes Tdh1, 2, and 3. In addition, the paper contains a section on benchmarking of the base editor in yeast, where it has not been used before.

Strengths and weaknesses of the paper:<br /> The authors establish the BE3 base editor as a screening tool in S. cerevisiae and very thoroughly benchmark its functionality for single edits and in different screening formats (fitness and FACS screening). This will be very beneficial for the yeast community.

The strategy established here allows measuring the effect of genetic perturbations on protein abundances in highly complex libraries. This complements capabilities for measuring effects of genetic perturbations on transcript levels, which is important as for some proteins mRNA and protein levels do not correlate well. The ability to measure proteins directly therefore promises to close an important gap in determining all their regulatory inputs. The strategy is furthermore broadly applicable beyond the current study. All experimental procedures are very well described and plasmids and scripts are openly shared, maximizing utility for the community.

There is a good balance between global analyses aimed at characterizing properties of the regulatory network and more detailed analyses of interesting new regulatory relationships. Some of the key conclusions are further supported by additional experimental evidence, which includes re-making specific mutations and confirming their effects on protein levels by mass spectrometry.

The conclusions of the paper are mostly well supported, but I am missing some analyses on reproducibility and potential confounders and some of the data analysis steps should be clarified.

The paper starts on the premise that measuring protein levels will identify regulators and regulatory principles that would not be found by measuring transcripts, but since the findings are not discussed in light of studies looking at mRNA levels it is unclear how the current study extends knowledge regarding the regulatory inputs of each protein.

Specific comments regarding data analysis, reproducibility, confounders:<br /> The authors use the number of unique barcodes per guide RNA rather than barcode counts to determine fold-changes. For reliable fold changes the number of unique barcodes per gRNA should then ideally be in the 100s for each guide, is that the case? It would also be important to show the distribution of the number of barcodes per gRNA and their abundances determined from read counts. I could imagine that if the distribution of barcodes per gRNA or the abundance of these barcodes is highly skewed (particularly if there are many barcodes with only few reads) that could lead to spurious differences in unique barcode number between the high and low fluorescence pool. I imagine some skew is present as is normal in pooled library experiments. The fold-changes in the control pools could show whether spurious differences are a problem, but it is not clear to me if and how these controls are used in the protein screen.

I like the idea of using an additional barcode (plasmid barcode) to distinguish between different cells with the same gRNA - this would directly allow to assess variability and serve as a sort of replicate within replicate. However, this information is not leveraged in the analysis. It would be nice to see an analysis of how well the different plasmid barcodes tagging the same gRNA agree (for fitness and protein abundance), to show how reproducible and reliable the findings are.

From Fig 1 and previous research on base editors it is clear that mutation outcomes are often heterogeneous for the same gRNA and comprise a substantial fraction of wild-type alleles, alleles where only part of the Cs in the target window or where Cs outside the target window are edited, and non C-to-T edits. How does this reflect on the variability of phenotypic measurements, given that any barcode represents a genetically heterogeneous population of cells rather than a specific genotype? This would be important information for anyone planning to use the base editor in future.

How common are additional mutations in the genome of these cells and could they confound the measured effects? I can think of several sources of additional mutations, such as off-target editing, edits outside the target window, or when 2 gRNA plasmids are present in the same cell (both target windows obtain edits). Could some of these events explain the discrepancy in phenotype for two gRNAs that should make the same mutation (Fig S4)? Even though BE3 has been described in mammalian cells, an off-target analysis would be desirable as there can be substantial differences in off-target behavior between cell types and organisms.

In the protein screen normalization uses the total unique barcode counts. Does this efficiently correct for differences from sequencing (rather than total read counts or other methods)? It would be nice to see some replicate plots for the analysis of the fitness as well as the protein screen to be able to judge that.

In the main text the authors mention very high agreement between gRNAs introducing the same mutation but this is only based on 20 or so gRNA pairs; for many more pairs that introduce the same mutation only one reaches significance, and the correlation in their effects is lower (Fig S4). It would be better to reflect this in the text directly rather than exclusively in the supplementary information.

When the different gRNAs for a targeted gene are combined, instead of using an averaged measure of their effects the authors use the largest fold-change. This seems not ideal to me as it is sensitive to outliers (experimental error or background mutations present in that strain).

Phenotyping is performed directly after editing, when the base editor is still present in the cells and could still interact with target sites. I could imagine this could lead to reduced levels of the proteins targeted for mutagenesis as it could act like a CRISPRi transcriptional roadblock. Could this enhance some of the effects or alter them in case of some missense mutations?

I feel that the main text does not reflect the actual editing efficiency very well (the main numbers I noticed were 95% C to T conversion and 89% of these occurring in a specific window). More informative for interpreting the results would be to know what fraction of the alleles show an edit (vs wild-type) and how many show the 'complete' edit (as the authors assume 100% of the genotypes generated by a gRNA to be conversion of all Cs to Ts in the target window). It would be important to state in the main text how variable this is for different gRNAs and what the typical purity of editing outcomes is.

Comments regarding findings:<br /> It would be nice to see a comparison of the results to the effects of ~1500 yeast gene knockouts on cellular transcriptomes (https://doi.org/10.1016/j.cell.2014.02.054). This would show where the current study extends established knowledge regarding the regulatory inputs of each protein and highlight the importance of directly measuring protein levels. This would be particularly interesting for proteins whose abundance cannot be predicted well from mRNA abundance.

The finding that genes that affect only one or two proteins are enriched for roles in transcriptional regulation could be a consequence of 'only' looking at 10 proteins rather than a globally valid conclusion. Particularly as the 10 proteins were selected for diverse functions that are subject to distinct regulatory cascades. ('only' because I appreciate this was a lot of work.)

Reviewer #3 (Public Review):

This manuscript wades into a research area that has risen to prominence during the COVID-19 pandemic, namely the estimation of time-varying quantities to describe transmission dynamics, based on case data collected in a given location. The authors focus on the interesting and challenging setting of low-incidence periods that arise after epidemic waves, when local spread of the virus has been contained, but new cases continue to be seeded by travelers and local spread potential can change as control measures are relaxed. There are important questions that arise in this context, such as when it is safe to declare the pathogen locally eliminated, and how to detect a flare-up quickly enough to stamp it out.

The authors propose a new framework, made up of a smoothed estimate of the local reproductive number, R, and another quantity they call Z, which is a measure of confidence that the local epidemic has been eliminated. They apply this framework to three public data sets of COVID-19 case reports (in New Zealand, Hong Kong and Victoria, Australia), each spanning multiple waves of infections interspersed with quieter periods when most cases arise from importation. They show how the smoothed R estimates align with the reported case data, and accurately capture periods of supercritical (R>1, so epidemics can take off) and subcritical (R<1, so epidemics wane) local transmission. They also show how the Z metric fluctuates through time, rising to near 100% at a few points which correspond closely to official declarations of elimination in the respective settings. The authors draw some parallels between their inferred R and Z metrics and the changes in control policies on the ground. They also highlight a number of points where the R and Z metrics seem to anticipate changes in the epidemiology on the ground, which are interpreted as advance 'signals' or 'early-warning' of ensuing waves of cases. This interpretation seems to underlie the manuscript's overall framing in terms of 'early-warning signals' that can be used 'in real time'.

Taken at face value, these are exciting claims that could form the basis of a useful public health tool. However I was not convinced that the framework was actually making these predictions in real time, i.e. strictly prospectively using available data. The approach would still have value if applied retrospectively, particularly with regard to understanding the impact of interventions applied in each setting. To this end, a more formal analysis of the relation between control measures and the R and Z metrics would benefit the paper.

Strengths

The paper is exemplary in clearly delineating the roles of importation versus local transmission in shaping case incidence during these low-incidence periods. This is a crucial distinction in this context, which is too often blurred.

The authors also innovate by bringing a suite of Bayesian filtering and smoothing techniques to bear on inferring R from these data, with the goal of extracting the cleanest signal possible from the noisy data. These approaches are well contextualized relative to more standard techniques in epidemiology, and appear to bear fruit in terms of smooth and stable estimates. However, it is important to note that this manuscript is not the primary report on these methods; the authors have written up this work elsewhere (ref. 16) and it is not described with sufficient detail for this manuscript to stand alone.

It is an interesting and valuable idea to derive a metric (Z) that explicitly estimates the degree of confidence that the pathogen has been eliminated locally. Again, the present manuscript builds closely on prior work by the authors (ref. 15), with the innovation of blending the earlier theory with the new Bayesian smoothed estimates of R.

The selected data sets are perfectly suited to the problem at hand, and analyzing three parallel case studies allows for the behavior of the R-Z framework to be observed across contexts, which is valuable.

Weaknesses

As presented, the manuscript does not seem to show real-time early-warning signals, as I understand those terms. The forward-backward smoothing algorithms that form the backbone of the study estimate R_s (i.e. the value of R at time s) using case data from both before and after time s. That is, the algorithm relies on knowledge of future events and so it cannot be said to provide early warning in any practical sense. Similarly, the estimates of Z draw upon the same 'smoothing posterior' q_s, so they also rely on future knowledge. (I doubted my understanding of this point, given the strong framing of the manuscript and limited methodological details, but the full explication of the method in ref. 16 is quite clear that the 'filtering posterior' p_s is suitable for real-time estimation, but the smoothing approach is retrospective and requires knowledge of the full dataset.)

Viewed in this light, the 'early-warning signals' in the Results are actually just smoothing of the yet-to-occur case data, and thus sadly are much less exciting. It did seem too good to be true. If I have understood correctly, then the current framing of the work seems inappropriate - unless the authors can show that R and Z metrics estimated strictly from past data can provide reliable signals of coming events.

An alternative approach would be to use the framework as a retrospective tool, and use it to build quantitative understanding of the impact of control measures and to revisit the timing of declarations of elimination. Table 1 goes some distance toward describing the relationships between R and Z values and these policy shifts and announcements, but I struggled to pull much value from it. The table and associated text mostly come across as a series of anecdotes where R fell after NPIs were imposed, or rose again when local transmission occurred, but there is no analysis that takes advantage of the more refined estimates of R the authors have obtained with their smoothing approach. One issue is that the time windows included in the table are not contiguous, so all the vignettes feel disjointed.

As presented, while the concept of the Z metric is attractive, it was hard to discern any conclusions about how to make use of its value. In two of the datasets it rose to near 100%, which is a clear signal of elimination, but as noted these were periods when the WHO rule of thumb (28 days without new cases) was sufficient. At some other points, the authors emphasize the implications of Z dropping close to 0% (e.g. at the top of page 7: on July 5 in Hong Kong, Z  0% despite 21 days without local cases, and the authors highlight the contrast with the WHO rule). However these findings clearly arise from the smoothing of future data mentioned above (i.e. on July 5 in Hong Kong, R is rising to supercritical levels based on advance knowledge of the rapid rise in cases in the next few weeks). Thus these findings are not relevant to real-time decision support. Finally, there are several periods where Z fluctuates around 20-50% for reasons that are hard to discern (e.g. July in New Zealand, or April-May in Hong Kong). The authors write in that the Z score may exhibit a peak due to extinction of a particular viral lineage in Hong Kong, while other lineages continued to circulate. It is hard to grasp how this interpretation could apply, given the aggregated nature of the data; more evidence, or more refined arguments, are needed for this to be convincing.

In the big picture, the proposed framework is based on two quantities, R and Z, but there is no systematic analysis of how to interpret these two quantities jointly. It would be valuable, for instance, to see how these metrics perform on a two-dimensional R-Z phase space.

The authors acknowledge a number of assumptions and data requirements needed for this approach, as presented. These include perfect case observation, no asymptomatic transmission, perfect identification of imported versus locally infected cases, and no delays in reporting. The authors state that the excellent surveillance systems in their case-study locales minimize the impact of these assumptions, but the same cannot be said of most other places around the world. Digging deeper into the epidemiology, the distribution of serial intervals (a crucial input to the algorithm) is assumed to be invariant, even though it's been demonstrated to change when interventions are imposed (ref. 26), i.e. exactly the conditions of interest. Finally, superspreading is a prominent feature of the COVID-19 epidemiology (as nicely documented for Hong Kong, by one of the authors), but is not addressed by this model beyond allowing subtle fluctuations in R from day to day. Taken together, these strong assumptions and omissions raise questions about the real-world reliability of this framework. Given that the point of the manuscript is to develop more refined quantitative metrics, and that most of these assumptions will be violated in most settings, it would be valuable to demonstrate that the framework is robust to these violations.

Funny/rude prank in Windows 3.1

Author Response

Reviewer #2 (Public Review):

This manuscript is of interest for neuroscientists studying neural circuit mapping in late larval, juvenile, and adult zebrafish. The work adapts and refines methods for retrograde viral tracing in zebrafish, using conditional and transneuronal DNA cargoes, to gauge the structure, connectivity, and function of neurons. Overall, the methods described in the paper, combined with a suite of viral constructs that are made available, represent a practical advance for virus-based neural circuit mapping in zebrafish, although a few aspects of experimental design and data interpretation require strengthening.

This work provides methodological refinements and new constructs for retrograde neuronal tracing and functional testing of circuit elements in zebrafish. The authors of the manuscript put impressive efforts into developing methods that are compatible with currently available transgenic zebrafish lines. The authors developed the methods based on previously-described herpes simplex virus 1 (HSV1) and pseudotyped rabies virus (RV) with deleted G protein (RVΔG) as neuronal labeling tools. First, they explore and assess temperature's effect on viral infection efficiency. The results indicate that a temperature close to the viral host temperature is optimal. Second, they engineered HSV1 into the UAS system that either contained TVA or codon-optimized glycoprotein (zoSDG). In the lines that contained TVA, the authors delivered HSV1-UAS containing TVA to Gal4 zebrafish lines for specific cell type delivery. With Gal4/UAS, they expanded the tool to adapt the transgenic zebrafish system that is widely used. Because EnvA/TVA works as a system, they then inject EnvA- RVΔG to target neurons where TVA is prelocated for specific labeling. Because of the deleted glycoprotein in RV, the reproducibility of the virus was limited. Therefore, they showed another experiment that complemented the EnvA- RVΔG by co-injection of the HSV1 containing zoSDG (HSV1[UAS:zoSADG]) as a helper virus to assist RVΔG in the transneuronal spread. Using the resulting retrograde migration of RV, the authors visualized the firstorder upstream connections labeled by HSV1-TVA+ neurons. Appropriate for a methodological paper, the function of the viruses are well described and their properties are well documented. In some cases, however, supporting data are thin or anecdotal, and do not always sufficiently support the manuscript's claims and conclusions. Further data, more nuanced interpretations, and/or more circumspect discussion points are needed to address these concerns.

Strengths:

1) HSV1 contains double-stranded DNA that can incorporate into the genome without using a complicated process to increase replication efficiency.

2) Specific gene targeting with the EnvA-TVA system increases accuracy during gene delivery. The expanded toolkit enhances the targeting strategy to include a diversity of useful constructs for the structural and functional assessment of neural circuits.

3) By making their toolbox compatible with the Gal4/UAS system, the authors leverage a large collection of Gal4 lines already available to the zebrafish community.

4) The toolbox for virus-based circuit mapping is relatively immature in the zebrafish model. The methods and reagents introduced here complement the current anterograde tracing using VSV. They also fill a gap in viral tracing for circuit mapping in adult zebrafish, as the immune system in juveniles and adults tended to reduce the viral spread efficiency using other approaches.

Weaknesses:

1. One of the major concerns of using this method is temperature increase. In zebrafish, temperature increase has been used as a heat stressor and is known to accelerate and facilitate development at larvae stage also cause lethality. Because of this accelerated development, the neurons labeled with HSV1 under heated conditions might not be the consequence of efficient virus infection, but rather a byproduct of faster migration and differentiation of neurons and other cells. Although the authors stated that adult zebrafish could tolerate higher temperatures (see item 5, below), this is not the normal condition for mapping circuits function, and the virus, as indicated in the manuscript, is also used in larvae. Further justification will be required to convince the audience that the use of high temperatures is generally adaptable, including for mapping circuits involved in other circuits. This is especially a concern for the HPA, because of the challenges in distinguishing the stress is from HSV1-induced oxidative stress from heat-induced neural stress.

The reviewer raises the possibility that increased expression after injection of HSV1 at higher temperatures may reflect increased proliferation (accelerated development) rather than increased infection efficiency. This scenario implies that a substantial fraction of labeled neurons were infected as progenitors, which then divided and differentiated into neurons. This possibility, although formally possible, appears extremely unlikely for the following reasons.

1. The difference in the number of labeled neurons is very large. If this difference were due to a difference in the speed of development, there should be an enormous difference in (brain) size between fish kept at different temperatures. If present, this size difference should be easily observable, at least in larvae. However, we did not observe an obvious size difference.

2. Viral infection was studied primarily in adult fish where neurogenesis still occurs, but at low rates compared to development. Nevertheless, the difference in labeled neurons was very large.

3. Many labeled neurons showed elaborated morphologies and long-range projections. It appears very unlikely that such neurons and their projections can arise by differentiation from precursors within the given incubation time.

4. The quantification in Fig. 1C was performed specifically for neurons with long-range projections in adult fish. The virus was injected into the OB while the neuronal somata were located in the dorsal telencephalon. If these neurons arose from precursors at the injection site, it would have to be postulated that these precursors migrated to the dorsal telencephalon, differentiated into neurons, and developed projections back to the OB. It is extremely unlikely that this can occur within the time of incubation. Moreover, there is no biological evidence for the migration of neuronal precursors or differentiating neurons from the OB to the dorsal telencephalon.

To further confirm that a speed-up of development cannot account for the observed difference in labeling we performed another variation of the experiment shown in Fig. 1: adult fish injected with HSV1[LTCMV:DsRed] into the OB were first kept at 36 deg for 3 days and then kept at 26 deg for 3 days before analysis (“36→26”). In these fish, DsRed expression in dorsal telencephalic neurons was indistinguishable from fish that were kept at 36 deg for the full period of 6 days. Fish that underwent the opposite temperature shift (“26→36”), in contrast, did not express DsRed in dorsal telencephalic neurons (Fig. 1C), despite the fact that they spent the same amount of time at each temperature.

Hence, the time at increased temperature per se cannot account for the difference in expression, indicating that temperature affects the process of infection. The new results have now been integrated into Fig. 1.

We cannot rule out that the temperature change affects stress levels and the HPA axis. However, as discussed in more detail below, swimming behavior was almost unchanged and obvious signs of stress were not observed. Moreover, please note that the temperature change can be restricted to the time around the virus injection, while any effects on behavior or neural activity will typically be examined several days later. Hence, effects of transgene expression will usually be evaluated at the standard laboratory temperature, long after the temperature change and the injection procedure.

2) HSV1 infects various cell types, not limited to neurons. The authors in the manuscript mentioned the high infection rate of cells. They did not categorize whether all infected cells were neurons or mixed neurons and glia. The authors briefly mention glia in the RNA sequencing data, but knowing the cell types and location is critical for circuit mapping. In Figure S2A-D, it seems that some of the cells around the midline could be radial glia. Cell migration from the midline is abundant, with radial-glia at the early stage guiding neurons from the ventricular zone to the mantle regions. How do authors ensure that the increased infection at higher temperatures does not include glia with the elevated immune response?

We do not claim that HSV1 infects only neurons. Indeed, HSV1 probably also infects glia, and the cells labeled in Fig. S2 are likely to include radial glia. However, this is not necessarily a disadvantage as additional specificity can be created by methods such as the Gal4 system. In fact, enhancing cell type specificity was a main motivation to combine HSV1 with the Gal4 system. A broad selectivity of the virus itself may then actually be considered an advantage because it allows for targeting of a broad spectrum of possible cell types. For example, HSV1 in combination with a transgenic line expressing Gal4 in glia (e.g., Tg[gfap:Gal4]) may be used to specifically interrogate glia cells if desired. We now discuss this issue of cell type specificity more specifically in the revised manuscript (ln 103-107; ln 339ff).

3)One limitation with HSV1 is that it resides inside neurons for an unpredictable length of time before expression, which increases the latency for induction of TVA. This extended latency could reduce sample size or lead to missed temporal windows. This caveat should be discussed.

We agree that the delay between HSV1 injection and transgene (TVA) expression may, in principle, decrease the efficiency of Rabies infection and retrograde tracing. We therefore performed a set of experiments in which we injected the Rabies virus 2 or 4 days after the HSV1. However, we observed lower, rather than higher, rates of Rabies infection, possibly because the sites of the two injections were not precisely identical. Hence, the advantage of staggered injections, if any, appears to be offset by variability in the location of injections, at least in our hands. Moreover, previous applications in rodents also reported high efficiency of Rabies infection when the Rabies virus was applied at the same time as the TVA expression construct (Vélez-Fort et al. 2014; Wertz et al. 2015). We now show results in Figure 4 – figure supplement 1 and discuss these issues briefly in the revised manuscript (ln 271-274; ln 677ff).

4). In the manuscript, to achieve transneuronal labeling, the fish were exposed to three viruses across two injections. The approach also includes exposure to chronicle heat, selection of TVA+ neurons from the first round of injection, and long periods of incubation between steps in the protocol. This is both labor-intense and potentially challenging for the animals' health and survival. Because the rates of lethality and poor health are not quantified for times after the first injection, and because the efficiency of the labelling approach (assessed at the animal level) are not reported, it is difficult to judge whether the approach is efficient enough for experimental work, where a large n of animals will be necessary for multiple treatments. This is particularly the case for phenotyping where mutant lines may be predisposed to adverse effects from heat or other manipulations and interventions. The manuscript would ideally show the number of fish that 1) were injected, 2) were infected with the virus, 3) survived until the timepoint for data collection, and 4) yielded publishable data. The possible limitations for studying mutants, especially those susceptible to heat and infection, should be discussed.

We agree that more information on the success rate and survival rates is desired. Previously, we had not explicitly reported survival rates because these were very high, and we apologize for not mentioning this explicitly. In the revised manuscript, we have now addressed this issue more specifically.

Please note that all fish used in experiments are represented by individual data points in the figures (except for a very low number of fish that did not survive the injection); no fish were excluded from the analysis. This is now pointed out explicitly in Methods (“Statistical analysis”). Hence, the data in the figures show directly how many fish were infected with the virus (point 2 above; 100% of injected fish) and how many neurons were labled in each fish. In all fish, images were acquired and the number of labeled neurons was quantified, implying that all fish yielded “publishable data” (point 4 above).

The survival rate (points 1 and 3 above) was very close to 100% in adult fish, and very few fish were lost during the injection. This has now been quantified systematically for all experimental conditions. We directly compared the survival of fish that were not injected, injected with buffer, and injected with virus, either at standard laboratory temperature (typically 26 deg for adults, 28.5 deg for larvae) or at elevated temperature (36 or 35 deg, respectively). The results are shown in Figure 1 – figure supplement 1.

In adult fish, survival rates were 100% under all conditions after single injections of HSV1 viruses. In larvae, some mortality was observed under control conditions that was slightly enhanced at elevated temperatures. We speculate that this is an indirect effect because larvae were kept in petri dishes in stagnant medium and water quality degrades more rapidly at higher temperatures. In any case, survival rates one week after injection were still relatively high (~50%). Moreover, for the first 3 days, survival rates were >90%. This appears particularly relevant because two or three days of exposure to high temperature are sufficient to achieve efficient expression. Survival rates were still 80 – 90% after two injections of HSV1 or after injections of rabies virus. Hence, the temperature shift should be compatible with a broad spectrum of practical applications. No effect of the HSV1 itself was detected on survival rates.

5) The current videos do not provide a rigorous demonstration that animals routinely tolerate elevated temperatures or infection (S Movies 1-3). Rates of survival for these cohorts and quantification of their swim behavior (such as distance travelled) with statistics would be more convincing. This criticism applies even more strongly to the single video of a sick fish (S Movie 4), which the authors use to support a claim of a targeted circuit manipulation using TeTx.

We have now quantified swimming behavior using two approaches. First, we compared the mean swimming speed between the six experimental groups used to determine effects of temperature and HSV1 on survival rates. Swimming was quantified at 27 deg after keeping fish at either 27 deg or at 36 deg for seven days. No significant difference in swimming behavior was observed (Figure 1 – figure supplement 2).

In addition, we quantified swimming behavior of fish at room temperature (25 – 26 deg) or 36 deg. Fish were kept in groups of five and individual fish were tracked using a machine learning-based tracking software (DeepLabCut). This allowed us to quantify different behavioral components. We found that mean swimming speed was higher at 36 deg and fish stayed slightly higher in the water column. However, social distance and the visual appearance of swimming were not obviously different. Swimming speed was normal again when fish were returned to normal temperature after seven days at 36 deg. These data are now shown in Figure 1 – figure supplement 2A,B.

6) FACS sorting and transcriptomics is a very complex and not wholly informative approach for judging stress at the cellular and organismal level. First, stress level is best assessed with high temporal resolution and best measured through blood or whole body (for larvae) cortisol measurements. Second, it is best to judge stress circuits in zebrafish in the diencephalon-mesencephalon, for the HPA. Cellular stress could best be measured with IHC for oxidative stress in infected cells and for apoptotic cells in the wake of infections. Taking measurements from OB neurons, with RNA sequencing that followed the elimination of dead cells during tissue disassociation and cell sorting, could have missed elements of the stress process. The sequencing result from only live cells in the OB may not provide the most reliable evidence.

We believe that there is a misunderstanding here. We did not analyze stress at the organismal level or activation of the HPA axis. In fact, we compared cells collected from the same individuals, which rules out any differences in organismal stress levels between samples. Organismal stress is not a topic of this study; addressing this is clearly beyond the scope of this study.

The transcriptomics experiments were specifically designed to examine cellular stress caused by Rabies infection. We agree that the transcriptomics approach has limitations but we feel that the data nevertheless contain valuable information. Together with other findings (morphology, calcium imaging), they support the conclusion that infection by the Rabies virus (in the absence of G) does not cause excessive cellular toxicity on the timescales of our experiments, consistent with results from other species. We agree that it is possible that the Rabies virus has more subtle effects on cellular stress levels (or immune responses) but a detailed analysis of such effects is beyond the scope of the present study. This is now discussed explicity (Results: ln 224-228; Discussion: ln 372-375). It is also possible that toxicity would occur on longer timescales. This may be expected based on findings in rodents but still leaves a broad time window for anatomical and functional experiments. This is now discussed explicity (ln 378-381).

7) The down-regulation in stress markers needs further discussion. Under chronic stress of heat exposure, exacerbation of HPA axis function could reduce glucocorticoids.

Please note that control and infected cells were from the same animals. The temperature regime can therefore not explain the differences in gene expression. Please also note that animals were not exposed to elevated temperature for days prior to cell collection.

Please also note that the down-regulation of genes was broad, affecting not only stress-related genes. Indeed, stress-related genes were not downregulated more frequently than other genes. Gene groups that were down-regulated most frequently are associated with immune responses. We therefore conclude that the downregulation of genes does not specifically reflect a stress response, and we speculate that it may reflect a general immune-related response. However, this is very hypothetical, and further studies are needed to understand the processes behind the observed pattern of gene regulation.

This is now stated clearly in the revised text (ln 224-228; ln 372-375).

8) Although it cannot be addressed for larvae, it is critical to report the sex ratio for your adults, since hormones affect stress and circuits formation.

Adult fish of both sexes in approximately a 50:50 ratio were used to ensure that there is no sexdependent bias in the data. However, the exact sex ratio in each experiment has not been recorded. This is now stated explicitly in Methods.

Reviewer #3 (Public Review):

Satou et al. report a viral toolbox by:

1) Inventing a novel way through temperature-dependence of HSV1-mediated gene expression for adult and larval zebrafish;

2) Employing Gal4/UAS system to achieve cell types specific expression in this model;

3) Combining the modified rabies viruses and HSV1 for transneuronal tracing of neural circuits in zebrafish that is kept in a higher temperature environment.

This toolbox in the manuscript will be of great interest to the neuroscience field when they are using zebrafish as a model.

The strength is these novel methods will offer more experimental opportunities and will facilitate more exciting basic scientific discoveries. However, some concerns still exist as below:

1) What's the mechanism of temperature-dependence expressions with these HSV1 and rabies virus in this study? At least the authors should discuss it. Have the authors done experiments like this: after getting enough gene expression from these viruses when maintaining these fishes in 35-37 degree, bring them back to normal temperature as they usually live to see what happen? Does this higher temperature help the fish brain cells get infected with more viral particles or just help increase the expression level? Or does just the higher temperature help produce more proteins?

The question raised by the reviewer is indeed interesting. We agree that it would be useful to know whether host-like temperature enhances the entry of the virus into the host cell (infection), viral replication/protein synthesis, or both. In the original manuscript, we reported results from a first experiment to address this question. In this experiment, we injected HSV1 at 26 deg and then increased temperature to 36 deg 3 days after infection (“2636”). This protocol yielded low expression. We have now also performed the reverse temperature shift (“3626”), as suggested by the reviewer. This protocol yielded high expression, comparable to the expression observed when fish were kept at 36 deg throughout (see Fig. 1). Together, these results suggest that temperature affected primarily the infection. However, additional, more advanced analyses are required to resolve to what extent temperature affects viral infection and viral replication/protein synthesis. This is now discussed explicitly (ln 332ff).

2) The authors should address or discuss more whether the higher temperature affects these fishes' brain activity? The reason is if someone will use this method for a most important experiment like GCaMP7s calcium imaging, in order to get good expression with these viruses that authors described in the manuscript they should raise the temperature but they have no idea about whether these higher temperatures affect the behavior or brain activity in some special brain regions they are interested in.

Please note that, in most applications, the temperature is increased only transiently around the time of injection for two or three days. Thereafter, fish can be transferred back to normal laboratory temperature without compromising transgene expression. Any follow-up experiments, e.g. analyses of behavior or neuronal activity, can therefore be performed at standard laboratory temperature, after fish were kept at this temperature for a few days. The temperature change should therefore have only minor, indirect effects on the results of behavioral or physiological experiments. We apologize if this was not evident and discuss this now explicitly (ln 334ff).

In addition, we have analyzed the swimming behavior of zebrafish in more detail at elevated temperatures (36 deg for adults; 35 deg for larvae) and observed only minor differences in swimming behavior. These results are now reported in Figure 1 – figure supplement 2A. Moreover, we compared swimming behavior between control fish (kept at standard laboratory temperature) and fish that underwent a transient temperature change to 36 deg for 7 days before the test. No significant difference in swimming behavior was observed between groups (Figure 1 – figure supplement 2B). We therefore conclude that no obvious effects of temperature are observed at least at the behavioral level.

Reviewer #2 (Public Review):

This manuscript is of interest for neuroscientists studying neural circuit mapping in late larval, juvenile, and adult zebrafish. The work adapts and refines methods for retrograde viral tracing in zebrafish, using conditional and transneuronal DNA cargoes, to gauge the structure, connectivity, and function of neurons. Overall, the methods described in the paper, combined with a suite of viral constructs that are made available, represent a practical advance for virus-based neural circuit mapping in zebrafish, although a few aspects of experimental design and data interpretation require strengthening.

This work provides methodological refinements and new constructs for retrograde neuronal tracing and functional testing of circuit elements in zebrafish. The authors of the manuscript put impressive efforts into developing methods that are compatible with currently available transgenic zebrafish lines. The authors developed the methods based on previously-described herpes simplex virus 1 (HSV1) and pseudotyped rabies virus (RV) with deleted G protein (RVΔG) as neuronal labeling tools. First, they explore and assess temperature's effect on viral infection efficiency. The results indicate that a temperature close to the viral host temperature is optimal. Second, they engineered HSV1 into the UAS system that either contained TVA or codon-optimized glycoprotein (zoSDG). In the lines that contained TVA, the authors delivered HSV1-UAS containing TVA to Gal4 zebrafish lines for specific cell type delivery. With Gal4/UAS, they expanded the tool to adapt the transgenic zebrafish system that is widely used. Because EnvA/TVA works as a system, they then inject EnvA- RVΔG to target neurons where TVA is prelocated for specific labeling. Because of the deleted glycoprotein in RV, the reproducibility of the virus was limited. Therefore, they showed another experiment that complemented the EnvA- RVΔG by co-injection of the HSV1 containing zoSDG (HSV1[UAS:zoSADG]) as a helper virus to assist RVΔG in the transneuronal spread. Using the resulting retrograde migration of RV, the authors visualized the first-order upstream connections labeled by HSV1-TVA+ neurons. Appropriate for a methodological paper, the function of the viruses are well described and their properties are well documented. In some cases, however, supporting data are thin or anecdotal, and do not always sufficiently support the manuscript's claims and conclusions. Further data, more nuanced interpretations, and/or more circumspect discussion points are needed to address these concerns.

Strengths:

1. HSV1 contains double-stranded DNA that can incorporate into the genome without using a complicated process to increase replication efficiency.<br /> 2. Specific gene targeting with the EnvA-TVA system increases accuracy during gene delivery. The expanded toolkit enhances the targeting strategy to include a diversity of useful constructs for the structural and functional assessment of neural circuits.<br /> 3. By making their toolbox compatible with the Gal4/UAS system, the authors leverage a large collection of Gal4 lines already available to the zebrafish community.<br /> 4. The toolbox for virus-based circuit mapping is relatively immature in the zebrafish model. The methods and reagents introduced here complement the current anterograde tracing using VSV. They also fill a gap in viral tracing for circuit mapping in adult zebrafish, as the immune system in juveniles and adults tended to reduce the viral spread efficiency using other approaches.

Weaknesses:

1. One of the major concerns of using this method is temperature increase. In zebrafish, temperature increase has been used as a heat stressor and is known to accelerate and facilitate development at larvae stage also cause lethality. Because of this accelerated development, the neurons labeled with HSV1 under heated conditions might not be the consequence of efficient virus infection, but rather a byproduct of faster migration and differentiation of neurons and other cells. Although the authors stated that adult zebrafish could tolerate higher temperatures (see item 5, below), this is not the normal condition for mapping circuits function, and the virus, as indicated in the manuscript, is also used in larvae. Further justification will be required to convince the audience that the use of high temperatures is generally adaptable, including for mapping circuits involved in other circuits. This is especially a concern for the HPA, because of the challenges in distinguishing the stress is from HSV1-induced oxidative stress from heat-induced neural stress.

2. HSV1 infects various cell types, not limited to neurons. The authors in the manuscript mentioned the high infection rate of cells. They did not categorize whether all infected cells were neurons or mixed neurons and glia. The authors briefly mention glia in the RNA sequencing data, but knowing the cell types and location is critical for circuit mapping. In Figure S2A-D, it seems that some of the cells around the midline could be radial glia. Cell migration from the midline is abundant, with radial-glia at the early stage guiding neurons from the ventricular zone to the mantle regions. How do authors ensure that the increased infection at higher temperatures does not include glia with the elevated immune response?

3. One limitation with HSV1 is that it resides inside neurons for an unpredictable length of time before expression, which increases the latency for induction of TVA. This extended latency could reduce sample size or lead to missed temporal windows. This caveat should be discussed.

4. In the manuscript, to achieve transneuronal labeling, the fish were exposed to three viruses across two injections. The approach also includes exposure to chronicle heat, selection of TVA+ neurons from the first round of injection, and long periods of incubation between steps in the protocol. This is both labor-intense and potentially challenging for the animals' health and survival. Because the rates of lethality and poor health are not quantified for times after the first injection, and because the efficiency of the labelling approach (assessed at the animal level) are not reported, it is difficult to judge whether the approach is efficient enough for experimental work, where a large n of animals will be necessary for multiple treatments. This is particularly the case for phenotyping where mutant lines may be predisposed to adverse effects from heat or other manipulations and interventions. The manuscript would ideally show the number of fish that 1) were injected, 2) were infected with the virus, 3) survived until the timepoint for data collection, and 4) yielded publishable data. The possible limitations for studying mutants, especially those susceptible to heat and infection, should be discussed.

5. The current videos do not provide a rigorous demonstration that animals routinely tolerate elevated temperatures or infection (S Movies 1-3). Rates of survival for these cohorts and quantification of their swim behavior (such as distance travelled) with statistics would be more convincing. This criticism applies even more strongly to the single video of a sick fish (S Movie 4), which the authors use to support a claim of a targeted circuit manipulation using TeTx.

6. FACS sorting and transcriptomics is a very complex and not wholly informative approach for judging stress at the cellular and organismal level. First, stress level is best assessed with high temporal resolution and best measured through blood or whole body (for larvae) cortisol measurements. Second, it is best to judge stress circuits in zebrafish in the diencephalon-mesencephalon, for the HPA. Cellular stress could best be measured with IHC for oxidative stress in infected cells and for apoptotic cells in the wake of infections. Taking measurements from OB neurons, with RNA sequencing that followed the elimination of dead cells during tissue disassociation and cell sorting, could have missed elements of the stress process. The sequencing result from only live cells in the OB may not provide the most reliable evidence.

7. The down-regulation in stress markers needs further discussion. Under chronic stress of heat exposure, exacerbation of HPA axis function could reduce glucocorticoids.

8. Although it cannot be addressed for larvae, it is critical to report the sex ratio for your adults, since hormones affect stress and circuits formation.

I've used ABBY FineReader (best on Windows) and it was much better at correcting OCR than Adobe Acrobat. —Dana Conard

Я это предложение перестроила, не уверена, что смысл не испортила, надо проверить

Author Response

Reviewer #1 (Public Review):

The authors examined the relationships between humans' heartbeats and their ability to perceive objects using touch.

Strengths: This study is a large and sophisticated one, with great attention to detail and systematic analysis of the resulting data. The hypotheses are clear and the study was carried out well. The presentation of the data visually is very informative. With such a large and high-quality set of data, the conclusions that we can draw should be clear and strong.

Weaknesses: The main drawbacks for me were first, exactly how the data were analysed, and second that there seem to be too many results reported to get an overall view of what the study has found.

First, there are always a number of choices that researchers can make when analysing their data. Too many choices in fact. So we always need to see a consistent, principled, and transparent account of how those choices were made and what the effects on the data were. At present, I think this needs to be improved, partly in the justification of the analyses that were done; partly by re-doing some analyses and the presentation of results.

Second, I admit to being a little lost when trying to understand all of the analyses - why there were done, what choices were made, and what the findings were. In some cases, it felt a little bit like the analyses were decided on only quite late - after exploring the data. One clear way to address this would be to divide the main results into two kinds: confirmatory (those that the authors expected to do before the study was run), and exploratory (those that the authors decided to do only after seeing the data). This would be both good practice and would help to focus the reader on what are the most critical findings.

Achievements: I think the presentation of results needs to be strengthened before I can decide whether the aims are achieved.

Impact: This will also depend on the revision of the results.

We thank the Reviewer for these comments. In the original manuscript we thought we have been clear as to those analyses that were planned and those that were exploratory. The planned analyses are in keeping with the previous studies in the literature on which this study was based (Al et al. 2020; Al et al. 2021; Grund et al. 2021). The only exploratory analysis was the inclusion of touch variance as a co-variate. We had not expected that participants would differ so much in how long they held their touch.

Reviewer #2 (Public Review):

In this article, the authors set out to discover whether the cardiac cycle influences active tactile discrimination, to better understand the putative relationship between interoception rhythms and exteroceptive perception. While numerous articles have looked at these relationships in the passive domain, here the authors designed an innovative active sensing task to better understand the interaction of sensorimotor processes with the cardiac rhythm.

The authors report a series of consecutive analyses. In the first, they find that while active discriminative touch is not modulated by the cardiac cycle, non-discriminative touch is such that the start, median duration, and end time of touches are shifted forward along the cardiac cycle towards diastole. Next, the authors examined the proportion of total start and end touches within systole versus diastole and found that across both discrimination and control conditions, touch was roughly 10-25% more likely to terminate during diastole. Further, examining the median holding time, the authors found that touches initiated during systole were lengthened in duration, consistent with a perceptual inhibition by this phase. This last effect appeared to be greatest for the highest stimulus difficulty levels, further supporting the notion that some cardiac inhibition of sensory processing may be at stake. Finally, when examining physiological responses, the authors found that cardiac inter-beat intervals were lengthened during active touch, consistent with the hypothesis that the brain may exploit strategic cardiac deceleration to minimize inhibitory effects.

Overall, the key effects of the manuscript are fascinating and robust. A major strength of the approach here is the task itself, which utilizes a well-controlled stimulus with multiple levels of task difficulty, as well as an elegant positive control condition. This enabled the authors to look rigorously at difficulty and stimulus condition interactions with the cardiac phase. This clearly pays off in the analyses, as the authors are able to construct a more informative story about how precisely cardiac timing events modulate perception.

Statistically speaking, I found the overall approach to be rigorous and sound. The study is well powered for a psychophysical investigation of this nature, and the interpretation of results is based on robust effects in the presence of a strong positive control.

We thank the reviewer for these positive comments on the original version of this paper.

Reviewer #3 (Public Review):

The manuscript presents a carefully designed and well-controlled study on active tactile perception and its relationship to internal bodily rhythms - the cardiac cycle. This work builds on previous studies which also showed that active perception/voluntary actions occur in certain phases of the cardiac cycle, but the previous research failed to show/was not designed to show the significance of these synchronizations for perception or behaviour. To my knowledge, this is the first report that seems to experimentally show that active perception in the cardiac diastole leads to behavioural advantages - better tactile discrimination.

The manuscript itself is very clearly written, the introduction is concise but sufficient, while the results section is very well organised and I especially like how the authors guide the reader through the analysis and additional steps taken to understand the findings even better.

Yet, despite careful study design, effective visualisations, and elegantly constructed story, there are some analytical choices that, in my opinion, are not sufficiently justified or explained (e.g., selecting a diastolic window equal in length to the duration of systole, instead of using the whole duration of diastole). Such analytical decisions could have (at least some) effects on the obtained results and thus conclusions drawn.

We thank the Reviewer for these comments. The analyses referred to here were planned and specifically the choice of the windows for defining systole and diastole were identical to the studies in the literature on which this study was based (Al et al. 2020; Al et al. 2021).

one way to handle night shootings to use shadows and light

Reviewer #2 (Public Review):

I am confused about the nature of Poisson models. If I am correct, the Poisson(a+b) is the sum of the two Poisson(a) and Poisson(b), that is, Poisson(a+b) = Poisson(a) + Poisson(b). Then, the mixture and intermediate models are very similar, identical if a*lambda_A and (1-a)*lambda_B happen to be integer numbers.

It is unclear why the 'outside' model predicts responses outside the range if neurons were to linearly sum the A and B responses.

It is also unclear why the 'single' hypothesis would indicate a winner-take-all response. If I understand correctly, under this model, the response to A+B is either the rate A or B, but not the max between lambda_A and lambda_B. Also, this model could have given an extra free parameter to modulate its amplitude to the stimulus A+B.

The concept of "coarse population coding" can be misleading, as actual population coding can represent stimulus with quite good precision. The authors refer to the broad tuning of single cells, but this does not readily correspond to coarse population coding. This could be clarified.

As a complement to the correlation analysis, one could check whether, on a trial-by-trial basis, the neuronal response of a single neuron is closer to the A+B response average, or to either the A or B responses. This would clearly indicate that the response fluctuates between representing A or B, or simultaneously represents A+B. I am trying to understand why this is not one of the main analyses of the paper instead of the correlation analysis, which involves two neurons instead of one.

In the discussion about noise correlations, the recent papers Nogueira et al., J Neuroscience, 2020 and Kafashan et al, Nat Comm, 2021 could be cited. Also, noise correlations can also be made time-dependent, so the distinction between the temporal correlation hypotheses and noise correlations might not be fundamental.

It would be interesting to study the effect of contrast on the mixed responses. Is it reasonable to predict that with higher contrast the mixture responses would be more dominant than the single ones? This could be the case if the selection mechanism has a harder time suppressing one of the object responses. This would also predict that noise correlations will go down with higher contrast.

What is the time bin size used for the analysis? Would the results be the same if one focuses on the early time responses or on the late time responses? At least from the units shown in Fig. 2, it looks that there is always an object response that is delayed respect to the other, so it would seem interesting to test noise correlations in those two temporal windows.

Reviewer #3 (Public Review):

Wang. et al explore the relationship between birth order and connectivity of neurons that wire together in a specific circuit. Using a refined single-cell clonal technique, the authors generate embryonic clones to map the birth order of neurons that derive from distinct stem cell lineages yet contribute to the same circuit in the Drosophila ventral nerve cord. Wang et. al map neurons of this circuit to discrete developmental windows, or "temporal cohorts," and show that neurons belonging to early vs. late temporal cohorts have stereotyped morphologies, wiring patterns, and birth orders. They convincingly show that distinct stem cell lineages contribute to the output vs input neurons of the circuit and that the output neurons are born before the input neurons. As a result, the authors provide novel insights into the relationship between neurogenesis and circuit assembly.

Strengths<br /> The relationship between birth order and connectivity at a single-cell resolution is a valuable step forward in understanding how cells are wired together during development. By dissecting the circuit into its individual subtypes and working backwards to birthdate the neurons, Wang et al take an unbiased and effective approach to understand how cells involved in sensing vibrational stimuli assemble within the ventral nerve cord.

The temporal mapping of neurons within and between lineages is challenging and laborious work and the data presented is clear and convincing. The authors modified the existing ts-MARCM system with the addition of another recombinase to facilitate the visualization of clones at earlier stages of development; this technique will be of use to members of the fly community.

The authors effectively demonstrate how analysis of the connectome data can be used to infer multiple aspects of neuronal development, including whether neurons share a common neuroblast parent (clustering of cell bodies) and their relative birth order (comparative cortex-neurite length).

Weaknesses<br /> A major conclusion of the paper is that the output neurons in a circuit are made before the input neurons. However, the strength of this conclusion is weakened by the fact that ~34% of the interneuron inputs to the EL-early neurons remain unmapped. This includes six neurons that synapse onto the EL-early neurons over 10 times each. It is therefore likely that other lineages contribute neurons that synapse onto the EL-early neurons. Without knowing the relative birthdates of these neurons to the early-EL neurons, the output first-input second conclusion should be tempered.

More consideration/discussion should be given to the tTF windows that these cohorts are derived from. For example, it would be intriguing if the early-EL, Ladder and Basin neuronal cohorts are all derived from the same tTF window. This would suggest that wiring specificity within a circuit is driven by the tTFs.

試試看 conda install py-xgboost

Reviewer #3 (Public Review):

Zadbood and colleagues investigated the way key information used to update interpretations of events alter patterns of activity in the brain. This was cleverly done by the use of "The Sixth Sense," a film featuring a famous "twist ending," which fundamentally alters the way the events in the film are understood. Participants were assigned to three groups: (1) a Spoiled group, in which the twist was revealed at the outset, (2) a Twist group, who experienced the film as normal, and (3) a No-Twist group, in which the twist was removed. Participants were scanned while watching the movie and while performing cued recall of specific scenes. Verbal recall was scored based on recall success, and evidence for descriptive bias toward two ways of understanding the events (specifically, whether a particular character was or was not a ghost). Importantly, this allowed the authors to show that the Twist group updated their interpretation. The authors focused on regions of the Default Mode Network (DMN) based on prior studies showing responsiveness to naturalistic memory paradigms in these areas and analyzed the fMRI data using intersubject pattern similarity analysis. Regions of the DMN carried patterns indicative of story interpretation. That is, encoding similarity was greater between the Twist and No-Twist groups than in the Spoiled group, and retrieval similarity was greater between the Twist and Spoiled groups than in the No-Twist group. The Spoiled group also showed greater pattern similarity with the Twist group's recall than the No-Twist group's recall. The authors also report a weaker effect of greater pattern similarity between the Spoiled group's encoding and the Twist group's recall than between the Twist group's own encoding and recall. Together, the data all converge on the point that one's interpretation of an event is an important determinant of the way it is represented in the brain.

This is a really nice experiment, with straightforward predictions and analyses that support the claims being made. The results build directly on a prior study by this research group showing how interpretational differences in a narrative drive distinct neural representations (Yeshurun et al., 2017), but extend an understanding of how these interpretational differences might work retrospectively. I do not have any serious concerns or problems with the manuscript, the data, or the analyses. However I have a few points to raise that, if addressed, would make for a stronger paper in my opinion.

1) My most substantive comment is that I did not find the interpretive framework to be very clear with respect to the brain regions involved. The basic effects the authors report strongly support their claims, but the particular contributions to the field might be stronger if the interpretations could be made more strongly or more specifically. In other words: the DMN is involved in updating interpretations, but how should we now think about the role of the DMN and its constituent regions as a result of this study? There are a number of ideas briefly presented about what the DMN might be doing, but it just did not feel very coherent at times. I will break this down into a few more specific points:

While many of us would agree that the DMN is likely to be involved in the phenomena at hand, I did not find that the paper communicated the logic for singularly focusing on this subset of regions very compellingly. The authors note a few studies whose main results are found in DMN regions, but I think that this could stand to be unpacked in a more theoretically interesting way in the Introduction.

Relatedly, I found the summary/description of regional effects in the Discussion to be a bit unsatisfying. The various pattern similarity comparisons yielded results that were actually quite nonoverlapping among DMN regions, which was not really unpacked. To be clear, it is not a 'problem' that the regional effects varied from comparison to comparison, but I do think that a more theoretical exploration of what this could mean would strengthen the paper. To the authors' credit, they describe mPFC effects through the lens of schemas, but this stands in contrast to many other regions which do not receive much consideration.

Finally, although there is evidence that regions of the DMN act in a coordinated way under some circumstances, there is also ample evidence for distinct regional contributions to cognitive processes, memory being just one of them (e.g., Cooper & Ritchey, 2020; Robin & Moscovitch, 2017; Ranganath & Ritchey, 2012). The authors themselves introduce the idea of temporal receptive windows in a cortical hierarchy, and while DMN regions do appear to show slower temporal drift than sensory areas, those studies show regional differences in pattern stability across time even within DMN regions. Simply put, it is worth considering whether it is ideal to treat the DMN as a singular unit.

2) I think that some direct comparison to regions outside the DMN would speak to whether the DMN is truly unique in carrying the key representations being discussed here. I was reluctant to suggest this because I think that the authors are justified in expecting that DMN regions would show the effects in question. However, there really is no "null" comparison here wherein a set of regions not expected to show these effects (e.g., a somatosensory network, or the frontoparietal network) in fact do not show them. There are not really controls or key differences being hypothesized across different conditions or regions. Rather, we have a set of regions that may or may not show pattern similarity differences to varying degrees, which feels very exploratory. The inclusion of some principled control comparisons, etc. would bolster these findings. The authors do include a whole-brain analysis in Supplementary Figure 1, which indeed produced many DMN regions. However, notably, regions outside the DMN such as the primary visual cortex and mid-cingulate cortex appear to show significant effects (which, based on the color bar, might actually be stronger than effects seen in the DMN). Given the specificity of the language in the paper in terms of the DMN, I think that some direct regional or network-level comparison is needed.

3) If I understand correctly, the main analyses of the fMRI data were limited to across-group comparisons of "critical scenes" that were maximally affected by the twist at the end of the movie. In other words, the analyses focused on the scenes whose interpretation hinged on the "doctor" versus "ghost" interpretation. I would be interested in seeing a comparison of "critical" scenes directly against scenes where the interpretation did not change with the twist. This "critical" versus "non-critical" contrast would be a strong confirmatory analysis that could further bolster the authors' claims, but on the other hand, it would be interesting to know whether the overall story interpretation led to any differences in neural patterns assigned to scenes that would not be expected to depend on differences in interpretation. (As a final note, such a comparison might provide additional analytical leverage for exploring the effect described in Figure 3B, which did not survive correction for multiple comparisons.)

4) I appreciate the code being made available and that the neuroimaging data will be made available soon. I would also appreciate it if the authors made the movie stimulus and behavioral data available. The movie stimulus itself is of interest because it was edited down, and it would be nice for readers to be able to see which scenes were included.

To sum up, I think that this is a great experiment with a lot of strengths. The design is fairly clean (especially for a movie stimulus), the analyses are well reasoned, and the data are clear. The only weaknesses I would suggest addressing are with regards to how the DMN is being described and evaluated, and the communication of how this work informs the field on a theoretical level.

Author Response

Reviewer #1 (Public Review):

The authors sought to create a machine learning framework for analyzing video recordings of animal behavior, which is both efficient and runs in an unsupervised fashion. The authors construct Selfee from recent computational neural network codes. As the paper is methodsfocused, the key metrics for success would be (1) whether Selfee performs similarly or more accurately than existing methods, and more importantly (2) whether Selfee uncovers new behavioral features or dynamics otherwise missed by those existing methods.

Weaknesses:

Although the basic schematics of Selfee are laid out, and the code itself is available, I feel that material in between these two levels of description is somewhat lacking. Details of what other previously published machine learning code makes up Selfee, and how those parts work would be helpful. Some of this is in the methods section, but an expanded version aimed at a more general readership would be helpful.

Thanks for the suggestions. We expanded the paragraphs describing training objectives and AR-HMM analysis. We also revised Figure 2C for clarity, and we have added a new figure, Figure 6, to describe how our pipeline works in detail. We also added a detailed instructions for Selfee usage on our GitHub page.

*The paper highlights efficiency as an important aspect of machine learning analysis techniques in the introduction, but there is little follow up with this aspect.

Our model only had a more efficient training process compared with other self-supervised learning methods. We also found our model could perform zero-shot domain transfer, so training may not even be necessary. However, we did not mean that our model was superior in terms of data efficiency or inference speed. We have revised some of the claims in the Discussion.

*In comparing Selfee to other approaches, the paper uses DeepLabCut, but perhaps running other recent methods for more comprehensive comparison would be helpful as well.

We compare Selfee feature extraction with features from FlyTracker or JAABA, two widely used software. We also visualized the tracking results of SLEAP and FlyTracker in complement to the DeepLabCut experiment.

*Using Selfee to investigate courtship behavior and other interactions was nicely demonstrated. Running it on simpler data (say, videos of individual animals walking around or exploring a confined space) might more broadly establish the method's usefulness.

We used Selfee with open field test (OFT) of mice after chronic immobilization stress (CIS) treatment. We demonstrated that our pipeline from data preprocessing to all the data mining algorisms with this experiment, and the results were added to the last section of Results.

Reviewer #2 (Public Review):

Jia et al. present a CNN based tool named "Selfee" for unsupervised quantification of animal behavior that could be used for objectively analyzing animal behavior recorded in relatively simple setups commonly used by various neurobiology/ethology laboratories. This work is very relevant but has some serious unresolved issues for establishing credibility of the method.

Overall Strengths: Jia et al have leveraged a recent development "Simple Siamese CNNs" to work for behavioral segmentation. This is a terrific effort and theoretically very attractive.

Overall Weakness: Unfortunately, the data supporting the method is not as promising. It is also riddled with incomplete information and lack of rationale behind the experiments.

Specific points of concern:

1) No formal comparison with pre-existing methods like JAABA which would work on similar videos as Selfee.

We added some comparisons with JAABA and FlyTracker extracted features, and also visualized FlyTracker and SLEAP tracking results aside from DeepLabCut. This result is now in the new Table 1. To avoid tracking inaccuracy during intensive interactions and potential inappropriately tuned parameters, we used a peer-reviewed dataset focused on wing extension behavior only. Our results showed a competitive performance of Selfee as other methods.

2) For all Drosophila behavior experiments, I'm concerned about the control and test genetic background. Several studies have reported that social behaviors like courtship and aggression are highly visual and sensitive to genetic background and presence of "white" gene. The authors use Canton S (CS) flies as control data. Whereas it is unclear if any or all of the test genotypes have been crossed into this background. It would be helpful if authors provide genotype information for test flies.

We have added a detailed sheet about their genotype in this version. The genetic information of all animals can also be found on the Bloomington fly center by the IDs provided. In brief, five fly lines used in this work are in the CS background: CCHa2-R-RAGal4, CCHa2-R-RBGal4, Dop2RKO, DopEcRGal4 and Tdc2RO54. We did not back cross other flies into the CS background for three reasons. First, most mutant lines are compared with their appropriate control lines. For example, in the original Figure 3B (the new Figure 4B), for CCHa2-R-RBGal4 > Kir2.1 flies contained wildtype white gene, so the comparison with CS flies would not cause any problem. For TrhGal4 flies, they were in white background, and so were other lines that had no phenotype. At the same time, in the original Figure 3G to J (the new Figure 4G to J), we used w1118 as controls for TrhGal4 flies, which were all in mutated white background. Second, in the original Figure 4F and G (the new Figure 5F and G), we admitted that the comparison between NorpA36, in mutated white background, and CS flies was not very convincing. Nevertheless, the delayed dynamic of NorpA mutants was reported before, and our experiment was just a demonstration of the DTW algorithm. Lastly, our method focused on the methodology of animal behavior analysis, and original videos were provided for research replications. Therefore, even if the behavioral difference was due to genetic backgrounds, it would not affect the conclusion that our method could detect the difference

3) Utility of "anomaly score" rests on Fig 3 data. Authors write they screened "neurotransmitter-related mutants or neuron silenced lines" (lines 251-252). Yet Figure 3B lacks some of the most commonly occurring neurotransmitter mutants/neuron labeling lines (e.g. Acetelcholine, GABA, Dopamaine, instead there are some neurotransmitter receptor lines, but then again prominent ones are missing). This reduces the credibility of this data.

First of all, this paper did not intend to conduct new screening assays, rather we used pre-existed data in the lab to demonstrate the application of Selfee. Previous work in our lab focused on the homeostatic control of fly behaviors, so most listed lines used here were originally used to test the roles of neuropeptides or neurons nutrient and metabolism regulation, such as CCHarelated lines, a CNMa mutant, and Taotie neuron silenced flies. There were some other important genes that were not involved in this dataset. Some most common transmitters are not included for two reasons. First, common neurotransmitters usually have a very global and broad effect on animal behaviors, and even if there is any new discovery, it could be difficult to interpret the phenomenon due to a large number of disturbed neurons. Second, most mutants of those common neurotransmitters are not viable, for example, paleGal4 as a mutant for dopamine; Gad1A30 for GABA, and ChATl3 for acetylcholine. However, we did perform experiments on serotonin-related genes (SerT and Trh), octopamine-related genes (Tdc and Oamb), and some other viable dopamine receptor mutants.

4) The utility of AR-HMM following "Selfee" analysis rests on the IR76b mutant experiment (Fig4). This is the most perplexing experiment! There are so many receptors implicated in courtship and IR76b is definitely not among the most well-known. None of the citations for IR76b in this manuscript have anything to do with detection of female pheromones. IR76b is implicated in salt and amino acid sensation. The authors still call this "an extensively studies (co)receptor that is known to detect female pheromones" (lines310-311). Unsurprisingly the AR-HMM analysis doesn't find any difference in modules related to courtship. Unless I'm mistaken the premise for this experiment is wrong and hence not much weight should be given to its results.

We have removed the Ir76b results from the Results. The demonstration of AR-HMM was now done with a mouse open field assay.

Reviewer #3 (Public Review):

This paper is describing a machine learning method applied to videos of animals. The method requires very little pre-processing (end-to-end) such as image segmentation or background subtraction. The input images have three channels, mapping temporal information (liveframes). The architecture is based on tween deep neural networks (Siamese network) and does not require human annotated labels (unsupervised learning). However, labels can still be used if they are produced, as in this case, by the algorithm itself - self-supervised learning. This flavor of machine learning is reflected in the name of the method: "Selfee." The authors are convincingly applying the Selfee to several challenging animal behavior tasks which results in biologically relevant discoveries.

A significant advantage of unsupervised and self-supervised learning is twofold: 1) it allows for discovering new behaviors, and 2) it doesn't require human-produced labels.

In this case of self-supervised learning the features (meta-representations) are learned from two views of the same original image (live-frame), where one of the views is augmented in several different ways, with a hope to let the deep neural network (ResNet-50 architecture in this case) learn to ignore such augmentations, i.e. learn the meta-representations invariant to natural changes in the data similar to the augmentations. This is accomplished by utilizing a Siamese Convolutional Neural Network (CNN) with the ResNet-50 version as a backbone. Siamese networks are composed of tween deep nets, where each member of the pair is trying to predict the output of another. In applications such as face recognition they normally work in the supervised learning setting, by utilizing "triplets" containing "negative samples." These are the labels.

However, in the self-supervised setting, which "Selfee" is implementing, the negative samples are not required. Instead the same image (a positive sample) is viewed twice, as described above. Here the authors use the SimSiam core architecture described by Chen, X. & He, K (reference 29 in the paper). They add Cross-Level Discrimination (CLD) to the SimSiam core. Together these two components provide two Loss functions (Loss 1 and Loss 2). Both are critical for the extraction of useful features. In fact, removing the CLD causes major deterioration of the classification performance (Figure 2-figure supplement 5).

The authors demonstrate the utility of the Selfee by using the learned features (metarepresentations) for classification (supervised learning; with human annotation), discovering short-lasting new behaviors in flies by anomaly detection, long time-scale dynamics by ARHMM, and Dynamic Time Warping (DTW).

For the classification the authors use k-NN (flies) and LightGBM (mice) classifiers and they infer the labels from the Selfee embedding (for each frame), and the temporal context, using the time-windows of 21 frames and 81 frames, for k-NN classification and LightGBM classification, respectively. Accounting for the temporal context is especially important in mice (LightGBM classification) so the authors add additional windowed features, including frequency information. This is a neat approach. They quantify the classification performance by confusion matrices and compute the F1 for each.

Overall, I find these classification results compelling, but one general concern is the criticality of the CLD component for achieving any meaningful classification. I would suggest that the authors discuss in more depth why this component is so critical for the extraction of features (used in supervised classification) and compare their SimSiam architecture to other methods where the CLD component is implemented. In other words, to what degree is the SimSiam implementation an overkill? Could a simpler (and thus faster) method be used - with the CLD component - instead to achieve similar end-to-end classification? The answer would help illuminate the importance of the SimSiam architecture in Selfee.

We added more about the contribution of the CLD loss in the last paragraph of Siamese convolutional neural networks capture discriminative representations of animal posture, the second section of Results. Further optimization of neural network architectures was discussed in the Discussion section. As for why CLD is that important, there are two main reasons. First of all, all behavior photos are so similar that it is not very easy to distinguish them from each other. In the field of so-called self-supervised learning without negative samples, researchers use either batch normalization or similar operations to implicitly utilize negative samples within a minibatch. However, when all samples are quite similar, it might not be enough. CLD uses explicit clusters to utilize negative samples within a minibatch, in the word of the authors “Our key insight is that grouping could result from not just attraction, but also common repulsion”, so that provides more powerful discrimination. The second reason is what the author argued in the CLD paper, CLD is very powerful in processing long-tailed datasets. As shown in the original Figure 2—figure supplement 5 (the new Figure 3—figure supplement 5), behavior data are highly unbalanced. As explained in the CLD paper. CLD fights against long-tailed distribution from two aspects. One is that it scales up the importance of negative samples within a mini-batch from 1/B to 1/K by k-means; another is that cluster operation could relieve the imbalance between the tail and head classes within a mini-batch. Here I quote: “While the distribution of instances in a random mini-batch is long-tailed, it would be more flattened across classes after clustering.” It was also visualized in Fig5 of the CLD paper.

To the best of our knowledge, SimSiam is the simplest method that would work with CLD. In the original CLD paper, they combined CLD method with other popular frameworks including BYOL and Mocov2. However, those popular frameworks are more complicated than SimSiam networks. We have attempted to combine CLD with BarlowTwins but failed. As the author of CLD suggested on Github: “Hi, good to know that you are trying to combine CLD with BarLowTwins! My concern is also on the high feature dimension, which may cause the low clustering quality. Maybe it is necessary to have a projection layer to project the highdimensional feature space to a low-dimensional one.” In terms of speed, there are two major parts. For inference, only one branch is used, so the major contribution of efficiency comes from CNN backbone. In theory, light backbones like MobileNet would work, but ResNet50 is already fast enough on a model GPU. As for training, the major computational cost aside from the CNN backbone is from Siamese branches. Two branches, two times of computation. Nevertheless, CLD relied on this kind of structure, so even if the learning framework is simpler than Simsiam, it is not likely to achieve a faster training speed. As for other structures, I think this new instance learning framework (https://arxiv.org/abs/2201.10728) is possible to achieve a similar result with fewer data and in a shorter time. However, this powerful method could be used with CLD. We might try it in the future.

One potential issue with unsupervised/self-supervised learning is that it "discovers" new classes based, not on behavioral features but rather on some other, irrelevant, properties of the video, e.g. proximity to the edges, a particular camera angle, or a distortion. In supervised learning the algorithm learns the features that are invariant to such properties, because humanmade labels are used and humans are great at finding these invariant features. The authors do mention a potential limitation, related to this issue, in the Discussion ("mode splitting"). One way of getting around this issue, other than providing negative samples, is to use a very homogeneous environment (so that only invariance to orientation, translation, etc, needs to be accomplished). This has worked nicely, for example, with posture embedding (Berman, G. J., et al; reference 19 in the manuscript). Looking at the t-SNE plots in Figure 2 one must wonder how many of the "clusters" present there are the result of such learning of irrelevant (for behavior) features, i.e. how good is the generalization of the meta-representations. The authors should explore the behaviors found in different parts of the t-SNE maps and evaluate the effect of the irrelevant features on their distributions. For example, they may ask: to what extent does the distance of an animal from the nearest wall affect the position in the t-SNE map? It would be nice to see how various simple pre-processing steps might affect the t-SNE maps, as well as the classification performance. Some form of segmentation, even very crude, or simply background subtraction, could go a very long way towards improving the features learned by Selfee.

In the new Figure 3—figure supplement 1, the visualization demonstrates that our features contained a lot of physical information, including wing angles, animal distance and positions in the chamber. “Mode-split” can be partially explained by those features. We actually performed background subtraction and image crop for mice behaviors, where we found them useful.

The anomaly detection is used to find unusual short-lasting events during male-male interaction behavior (Figure 3). The method is explained clearly. The results show how Selfee discovered a mutant line with a particularly high anomaly score. The authors managed to identify this behavior as "brief tussle behavior mixed with copulation attempts." The anomaly detection analyses were also applied to discover another unusual phenotype (close body contact) in another mutant line. Both results are significant when compared to the control groups.

The authors then apply AR-HMM and DTW to study the time dynamics of courtship behavior. Here too, they discover two phenotypes with unusual courtship dynamics, one in an olfactory mutant, and another in flies where the mutation affects visual transduction. Both results are compelling.

The authors explain their usage of DTW clearly, but they should expand the description of the AR-HMM so that the reader doesn't have to study the original sources.

We expanded the section that talks about AR-HMM mechanisms.

Reviewer #3 (Public Review):

This paper is describing a machine learning method applied to videos of animals. The method requires very little pre-processing (end-to-end) such as image segmentation or background subtraction. The input images have three channels, mapping temporal information (live-frames). The architecture is based on tween deep neural networks (Siamese network) and does not require human annotated labels (unsupervised learning). However, labels can still be used if they are produced, as in this case, by the algorithm itself - self-supervised learning. This flavor of machine learning is reflected in the name of the method: "Selfee." The authors are convincingly applying the Selfee to several challenging animal behavior tasks which results in biologically relevant discoveries.

A significant advantage of unsupervised and self-supervised learning is twofold: 1) it allows for discovering new behaviors, and 2) it doesn't require human-produced labels.

In this case of self-supervised learning the features (meta-representations) are learned from two views of the same original image (live-frame), where one of the views is augmented in several different ways, with a hope to let the deep neural network (ResNet-50 architecture in this case) learn to ignore such augmentations, i.e. learn the meta-representations invariant to natural changes in the data similar to the augmentations. This is accomplished by utilizing a Siamese Convolutional Neural Network (CNN) with the ResNet-50 version as a backbone. Siamese networks are composed of tween deep nets, where each member of the pair is trying to predict the output of another. In applications such as face recognition they normally work in the supervised learning setting, by utilizing "triplets" containing "negative samples." These are the labels.

However, in the self-supervised setting, which "Selfee" is implementing, the negative samples are not required. Instead the same image (a positive sample) is viewed twice, as described above. Here the authors use the SimSiam core architecture described by Chen, X. & He, K (reference 29 in the paper). They add Cross-Level Discrimination (CLD) to the SimSiam core. Together these two components provide two Loss functions (Loss 1 and Loss 2). Both are critical for the extraction of useful features. In fact, removing the CLD causes major deterioration of the classification performance (Figure 2-figure supplement 5).

The authors demonstrate the utility of the Selfee by using the learned features (meta-representations) for classification (supervised learning; with human annotation), discovering short-lasting new behaviors in flies by anomaly detection, long time-scale dynamics by AR-HMM, and Dynamic Time Warping (DTW).

For the classification the authors use k-NN (flies) and LightGBM (mice) classifiers and they infer the labels from the Selfee embedding (for each frame), and the temporal context, using the time-windows of 21 frames and 81 frames, for k-NN classification and LightGBM classification, respectively. Accounting for the temporal context is especially important in mice (LightGBM classification) so the authors add additional windowed features, including frequency information. This is a neat approach. They quantify the classification performance by confusion matrices and compute the F1 for each.

Overall, I find these classification results compelling, but one general concern is the criticality of the CLD component for achieving any meaningful classification. I would suggest that the authors discuss in more depth why this component is so critical for the extraction of features (used in supervised classification) and compare their SimSiam architecture to other methods where the CLD component is implemented. In other words, to what degree is the SimSiam implementation an overkill? Could a simpler (and thus faster) method be used - with the CLD component - instead to achieve similar end-to-end classification? The answer would help illuminate the importance of the SimSiam architecture in Selfee.

One potential issue with unsupervised/self-supervised learning is that it "discovers" new classes based, not on behavioral features but rather on some other, irrelevant, properties of the video, e.g. proximity to the edges, a particular camera angle, or a distortion. In supervised learning the algorithm learns the features that are invariant to such properties, because human-made labels are used and humans are great at finding these invariant features. The authors do mention a potential limitation, related to this issue, in the Discussion ("mode splitting"). One way of getting around this issue, other than providing negative samples, is to use a very homogeneous environment (so that only invariance to orientation, translation, etc, needs to be accomplished). This has worked nicely, for example, with posture embedding (Berman, G. J., et al; reference 19 in the manuscript). Looking at the t-SNE plots in Figure 2 one must wonder how many of the "clusters" present there are the result of such learning of irrelevant (for behavior) features, i.e. how good is the generalization of the meta-representations. The authors should explore the behaviors found in different parts of the t-SNE maps and evaluate the effect of the irrelevant features on their distributions. For example, they may ask: to what extent does the distance of an animal from the nearest wall affect the position in the t-SNE map? It would be nice to see how various simple pre-processing steps might affect the t-SNE maps, as well as the classification performance. Some form of segmentation, even very crude, or simply background subtraction, could go a very long way towards improving the features learned by Selfee.

The anomaly detection is used to find unusual short-lasting events during male-male interaction behavior (Figure 3). The method is explained clearly. The results show how Selfee discovered a mutant line with a particularly high anomaly score. The authors managed to identify this behavior as "brief tussle behavior mixed with copulation attempts." The anomaly detection analyses were also applied to discover another unusual phenotype (close body contact) in another mutant line. Both results are significant when compared to the control groups.

The authors then apply AR-HMM and DTW to study the time dynamics of courtship behavior. Here too, they discover two phenotypes with unusual courtship dynamics, one in an olfactory mutant, and another in flies where the mutation affects visual transduction. Both results are compelling.

The authors explain their usage of DTW clearly, but they should expand the description of the AR-HMM so that the reader doesn't have to study the original sources.

Overall this paper introduces a potentially useful tool as well as several interesting biological results obtained by applying it to videos with very little pre-processing. Both, the method and the results are convincing.

(3 of 3)

Among the tools we'll be using this Summer is Slack. It will be our primary place of collaboration and communication. I highly suggest you download and install the app, as it is much more convenient that way.

Click the link below to join our class Slack Team. Slack will take you on a guided tour when you join - don't skip it! After the tour, go to the #introductions-welcome channel and follow the instructions in the final stage of our scavenger hunt.

STEME 4/5001 Summer 2022

Despite feeling uncomfortable, she stayed in the room as a result of John's convincing. Her perspective changed, and she began to appreciate the room and wallpaper and began to get rather fond of it.

I'm not sure this is a good sign

SciScore for 10.1101/2022.06.02.22275901: (What is this?)

Please note, not all rigor criteria are appropriate for all manuscripts.

Table 1: Rigor

Table 2: Resources

Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).

Results from LimitationRecognizer: We detected the following sentences addressing limitations in the study:

This is discordant to our MR results where higher genetically-predicted iron status is related to increased risk of sepsis and being hospitalized due to COVID-19, and could be attributed to differences in the epidemiological methods applied, such as residual confounding, but also limitations with the two-sample MR method used that is restricted to assess linear models (37). Few MR studies have explored iron status and risk of severe infections. An MR-study using iron related SNPs identified in the Genetics of Iron Status-consortia (38) found evidence that higher serum-iron, TSAT and ferritin were related to increased risk of sepsis (21). Using a more updated set of genetic instruments for iron status biomarkers, we replicated these findings for serum iron and TSAT, a tendency for TIBC, but not for ferritin. Another MR study found evidence of increased risk of skin and soft-tissue infections with higher serum iron levels (39). To the best of our knowledge, no previous study has conducted MR analysis to investigate the effect of iron status on incidence or outcome of COVID-19. Observational studies that have investigated iron status at the time of infection and found evidence of low iron status being a risk factor for a severe course of COVID-19 (12). Another study with COVID-19 patients compared to non-COVID-19 patients showed lower serum iron and TSAT levels in patients with COVID-19 independently of severity. Whereas COVID-19 patients defined as severe and critical had subst...

Results from TrialIdentifier: No clinical trial numbers were referenced.

Results from Barzooka: We did not find any issues relating to the usage of bar graphs.

Results from JetFighter: We did not find any issues relating to colormaps.

Results from rtransparent:

• Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
• Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
• No protocol registration statement was detected.

Results from scite Reference Check: We found no unreliable references.

About SciScore

SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

DOI: 10.12688/openreseurope.14784.1

Resource: Windows Azure (RRID:SCR_011880)

Curator: @scibot

SciCrunch record: RRID:SCR_011880

What is this?

<small><cite class='h-cite via'>ᔥ <span class='p-author h-card'>Boaz Barak</span> in Brian Conrad takes down the CMF – Windows On Theory (<time class='dt-published'>06/01/2022 21:35:06</time>)</cite></small>

"Darkness" is a word used to symbolize anxiety, suffering, or the general negativity within the world. This word is frequently repeated throughout the story, making it a motif.

Windows are also a motif in the story, and while one is not explicitly stated to be here the narrator seeing his own face would imply that is what he is looking at. Windows shed light on reality throughout the story, the window looking out into a dark tunnel in this case is used to symbolize denial.

Looking through windows is a common action in this story. Windows in art are usually used to symbolize change or hope. When one looks through the window, some light is shed on reality showing what is unattainable, sucking the hope out of those in Harlem.

lots of imagery

SciScore for 10.1101/2022.05.30.22275757: (What is this?)

Please note, not all rigor criteria are appropriate for all manuscripts.

Table 1: Rigor

Table 2: Resources

Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).

Results from LimitationRecognizer: We detected the following sentences addressing limitations in the study:

Advantages and limitations: In addition to important and actual findings related to the Covid-19 pandemic, some strengths as well as limitations should be discussed. The low response rate (7.4%) clearly limits representativeness of the results. On the other hand, the sample size became large enough for studying also small groups of participants and possible associations between various factors and functioning. The survey was not very long but it included also sensitive questions, which may have reduced individuals’ willingness to response. In client satisfaction surveys with no incentives, response rate often remains on the level of 10% or under (PeoplePulse, 2021). During the Covid-19 pandemic, the university students and personnel received several other surveys, thus it is probable that they were tired to response to a new survey. Additionally, the fact that this survey was carried out in May, when the term was near to end, may have affected low response rate. The study focused on people of university community, who do not represent the general population. On the other hand, the study sample represents a quite homogenous population, which faced equal and long-lasting Covid-19 lockdown with its consequences, when the differences in FUNCT between sub-groups of the sample, were mainly due to sub-group qualities than to different impacts of the pandemic. The question on gender included only four options (Female, Male, Other, I do not wish to tell) and all except reported female...

Results from TrialIdentifier: No clinical trial numbers were referenced.

Results from Barzooka: We did not find any issues relating to the usage of bar graphs.

Results from JetFighter: We did not find any issues relating to colormaps.

Results from rtransparent:

• Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
• Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
• Thank you for including a protocol registration statement.

Results from scite Reference Check: We found no unreliable references.

About SciScore

SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

SciScore for 10.1101/2022.05.27.22275696: (What is this?)

Please note, not all rigor criteria are appropriate for all manuscripts.

Table 1: Rigor

NIH rigor criteria are not applicable to paper type.

Table 2: Resources

Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).

Results from LimitationRecognizer: We detected the following sentences addressing limitations in the study:

Despite these limitations, a few conclusions can be safely drawn from our work. We show that over a third of the medical workforce in Maranhão and São Paulo was infected with COVID-19 in the first year of the pandemic, with a substantial loss of labour. This is consistent with the findings from smaller studies from Brazil(20) and other LMICs,(16,18,19) and therefore particularly relevant for those countries with a scarcity of healthcare resources, which will have been hit already particularly hard by the pandemic (35). The higher infection rate among Maranhão physicians was in contrast to lower population infection rates (see Tab.1). Our multivariate analysis confirmed that working in Maranhão was one of the most significant risk factors of physician infections in our cohort. The lower ratio of physicians per capita in Maranhão (1.1 per 1,000 in Maranhão Vs 3.2 per 1,000 in São Paulo)(27) may be a factor here, as during health emergencies a smaller workforce will necessarily engage in multiple functions and tasks across sectors, therefore increasing opportunities for infection. This is consistent with previous work(26) showing the differential impact of health system crises on unequal states in LMICs. If confirmed, such finding would be relevant for those studies forecasting effects of the pandemic on health workforces in different parts of the world (4) Younger age was associated with higher infection rates among physicians in both Brazilian states, which on the one hand con...

Results from TrialIdentifier: No clinical trial numbers were referenced.

Results from Barzooka: We did not find any issues relating to the usage of bar graphs.

Results from JetFighter: We did not find any issues relating to colormaps.

Results from rtransparent:

• Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
• Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
• Thank you for including a protocol registration statement.

Results from scite Reference Check: We found no unreliable references.

About SciScore

SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

0 FluxGarden

Согласно Левину и Левину [25] и Хименесу [12], мнемонические приемы могут оказывать положительное влияние не только на изучение языка, но и на любой тип обучения. Они требуют особого системного подхода. Метод локусов, например, включает в себя выбор различных элементов в знакомой среде, а затем создание мысленного образа для каждого элемента, который будет запоминаться вместе с каждым связанным словом [15], [26]–[29]. Недавние исследования показывают, что визуальная иммерсивная среда более эффективна для запоминания словарного запаса, чем традиционные методы. Эти исследования включают в себя виртуальную реальность [28], [29], дополненную реальность [15] и программу Memory Palace Beta [30]. Последнее представляет собой мультиплатформенное приложение, доступное для устройств Windows, Android и iOS, которое позволяет строить дворцы памяти с использованием метода локусов.

Один из используемых методов, метод локусов, был описан Цицероном еще в 55 г. до н.э. [13]. «Loci» на латыни — это множественное число от «locus», что означает место или положение. Эта концепция также известна под другими названиями, такими как Римская комната или Дворец памяти. Все они относятся к мнемонической стратегии, при которой знакомая среда активирует механизм ассоциации [14], облегчающий запоминание. Количество словарных единиц и количество времени, отведенное на запоминание в Римском дворце в пилотном исследовании, основано на результатах исследования, опубликованного Лархен Костухен, Дарлинг и Уйтман [15], которые показали, что запоминание 10 новых словарных единиц через 15 минут было значительно лучше при экспериментальном методе (карточки, связанные с 3D-моделями и просматриваемые в знакомой среде на мобильных устройствах), чем при обычном методе, состоящем из карточек с изображениями, представленными через приложение Quizlet для Windows, Android и iOS. Серьезная игра Roman Palace предназначена для разных возрастных групп, а также для разных индоевропейских языков. Пилотное исследование было сосредоточено на возрастной группе от 18 до 60 лет. Используемый язык был не настоящим, а скорее группой псевдослов, сгенерированных внешним инструментом. Эффективность сохранения словарного запаса измерялась в игре путем проверки перевода и правописания слов участниками. В игре также сохранялось время, затраченное на прохождение каждого уровня, непройденные слова в каждом тесте и полученные баллы. Несмотря на то, что мнемонические стратегии использовались со времен Цицерона [13] и часто упоминаются в академической литературе [13], [16]–[24], они не получили широкого применения в иммерсивных средах для конкретных целей (таких как словарь L2). приобретение), а научные публикации в этой области редки.

We cannot rely on software to determine the sector size of a disk. Must look at the specsheet.

do something that else hasn't done before

netscape on window a hole

never having to learn pointing windows

motivated to write software not on windows

编译器

SciScore for 10.1101/2022.05.12.22274991: (What is this?)

Please note, not all rigor criteria are appropriate for all manuscripts.

Table 1: Rigor

Table 2: Resources

Results from OddPub: Thank you for sharing your data.

Results from LimitationRecognizer: An explicit section about the limitations of the techniques employed in this study was not found. We encourage authors to address study limitations.

Results from TrialIdentifier: No clinical trial numbers were referenced.

Results from Barzooka: We did not find any issues relating to the usage of bar graphs.

Results from JetFighter: We did not find any issues relating to colormaps.

Results from rtransparent:

• Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
• Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
• No protocol registration statement was detected.

Results from scite Reference Check: We found no unreliable references.

About SciScore

SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

Note: This rebuttal was posted by the corresponding author to Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Reply to the reviewers

From the start, the authors would like to thank all the reviewers for their careful and constructive consideration of our manuscript. We have now made several changes to the paper and believe it to be better for the feedback.

Reviewer #1 (Evidence, reproducibility and clarity (Required)):

In this study, Rees et al. perform an RNA-seq circadian time course experiment in the recently formed allopolyploid wheat. Through comparisons with other circadian transcriptomic datasets in other species it appears that the period of rhythmic genes is much more variable in wheat with a shift to longer periods compared to the other species examined. Interestingly, by analyzing circadian parameters among expressed genes, they find evidence that this newly formed allopolyploid already shows signs of divergence in circadian traits among homoeologs. A thorough comparison with circadian regulated genes in Arabidopsis reveals overlap in phasing of genes involved in certain biological processes such as photosynthesis and light signaling whereas genes involved in starch metabolism were found to have different levels of rhythmicity and phasing. This dataset will be a great resource for the community and enable new predictions about the influence of polyploidy on the circadian control of important crop improvement traits and the circadian regulation of gene expression.

Major Comments

1. The results section starts with very little explanation of the experiment. It would help to provide a little more detail at the start of the results to explain the context for the experiment and what was done, when samples were collected and for how long. For the methods section, it isn't until line 650 that it is clearly stated that the sampling started at ZT0. It would be better to put this in the plant materials and growth condition section.

Thank you for highlighting the need for this context, we agree that the manuscript is improved by an introduction to the experiments. We have now included an “Experimental context” section in the results and have taken the opportunity to explain how the full 0-68h and 24-68h datasets are used within our analysis. Ln 74-82. We have also edited the Methods as suggested Ln 610-615.

The low proportion of circadian regulated genes is likely due to the very low cutoff for calling a gene expressed, especially when there are three days of repeated timepoints. If a gene is expressed across the time course it should have values above TPM 0 for at least 3 time points in order for it to be expressed each day. I'd also be suspicious of a gene with a TPM value less than 0.5. Comparing these types of numbers is always challenging due to the various cutoffs used. Along those lines, why was a different filtering scheme used for Arabidopsis (line 657)?

We completely agree that the proportion of genes described as rhythmic changes a great deal with the threshold at which you exclude low expression transcripts as well as the window over which measurements are taken and the q-value cut-off for rhythmicity. We performed an analysis to test the effects of applying a pre-filtering step to exclude low-expression genes and discuss our findings in Supplementary Note 1. Briefly, we removed genes with expression less than 0.1 TPM in six or more timepoints and again ran Metacycle to define numbers of rhythmic genes. Our results are discussed in Supplementary Note 1 and are presented in Supplementary Table 1. Regardless of the cut-offs applied, Arabidopsis and wheat data was treated identically, and our findings reported in the main results were consistent with those reported in the Supplementary analysis. Thank you for raising this point, as we have now improved our description of this analysis in the main text (Ln 92-95).

Regarding the different filtering schemes, the filtering mentioned by Reviewer 1 was applied to both Arabidopsis and wheat data for a stricter retention of rhythmic genes, as part of the pre-WGCNA clustering analysis. Filtering to retain genes with >0.5TPM across 3 timepoints was applied to reduce lowly expressed genes, that act as background 'noise' when defining clusters. We applied this across 3 timepoints rather than the WGCNA suggestion of 90% of samples - because the patterns of expression in our rhythmically filtered datasets were cyclical in nature.

In reference to the shortening of the period every day, this should be interpreted with caution. Period estimate of a single cycle are not very reliable and the SD for each day is around 3h so it is difficult to draw any conclusions about changes in period each day. One option would be to only include genes with an SD less than 1h or alternatively to remove the discussion surrounding the comparison of period across the three days and focus on the period results for the full 24h-68h window shown in 1b. While 2 days is better it is still not ideal for calling period; however, your first day will still have a strong diurnal driven pattern that will likely skew your circadian period.

Thank you for your comments. Our question here was to determine whether the mean period lengths of rhythmic transcripts in wheat were always immediately longer upon transfer to constant light, or whether they got progressively longer over time. Upon reading the reviewer’s comment, we realize that the explanation provided of how we conducted this analysis was misleading. Our approach was to take a 44h sliding window (almost 2 days) and measure period at 0-44h, 12-56h and 24-68h. We have now added the previously missing statistics that support our findings in the main text, and which hopefully show the significance of the period changes over time (supplementary note 2). One of the most surprising findings from this analysis was that the periods in the first window were the longest 28.61h (SD=3.421), suggesting that the diel (driven) oscillation had little impact upon immediate transfer to free run. Our interpretation is that the mean period initially lengthens trying to follow the missing dusk signal, before the free-running endogenous period asserts itself in later cycles (Ln 129-128).

Line 87-93: If the dusk cue is important for clock expression you would think this would be biased towards genes that peak later in the day or near dusk. This argument should be connected better to the period results discussed on lines 98-101.

Following on from our statement above, we have now combined our hypothesis for why wheat transcripts expressed at dusk have longer periods with the discussion about longer periods upon transfer to constant light. We agree that the two processes are likely to be connected and have now placed them together in Ln 129-128.

1. Lines 650-652 of the Methods mentions that one of the main interests was the response to transfer to L:L, but this isn't mentioned in the introduction and doesn't come up much in the Results section. Most of the expression comparisons are focused on the 24-68h window. It also isn't clearly explained why the first day in LL is still a diurnal cycle. This would be helpful for non-circadian readers who may wonder why the first day is not included in all the analyses.

We believe this point is now also addressed by the addition of an Experimental Context section in the results (Ln 74-82), in response to the reviewer’s previous comment.

1. The phase comparisons shown in Figure suppl 4 are confusing. Suppl. Note 3 states that the period from the 24-68h data window was used to establish the bins but then the phase is shown for 3 different windows for each column? When calculating the phase for each of those 3 windows which period was used as the denominator in the phase calculation? Was it the period that matches the window used to calculate phase? What does the plot look like if phase is called on the same window used to calculate period (24-68)? What method was used to call phase in Suppl. Fig 4? As shown in Suppl Fig. 3 the method can influence the phase distributions. The methods suggest that the phase was determined with Metacycle but then FFT and MESA were used to verify. What does this mean verify, were they adjusted if FFT/MESA didn't agree?

We agree that this Figure was unnecessarily complicated. We have now simplified Supplementary Figure 4 so that only the phases from 24-68h are presented. We have also clarified the legend to explain why we used FFT-NLLS to improve accuracy of Metacycle predictions.

It is difficult to interpret the value of the period and phase comparisons shown in Fig. 1b, c, e and f after the preceding section about how variable the period and phase is across days. It is also surprising that the full 3 days were used to calculate the circadian statistics considering the first day is still under diurnal control. Do the ratios remain the same if the statistics are performed only on the 24h-68h window? For consistency with the rest of the paper and avoid confusion it would be best to have all circadian parameters measured using the same time window (24h-68h).

Thank you for your comments, we can see how our logic in using the different data windows was not clear enough. As mentioned above, we have now explained the use of the full and shortened data windows in Experimental context section (Ln 74-82). Fig 1c is a comparison between different circadian datasets and as such we have only compared periods across 24-68h window. Similarly, Fig 1b is a global analysis of periods in rhythmic genes in comparison with Arabidopsis and so is again measured from 24-68h. We have now clarified this in the Figure legend for 1b.

For comparisons of homoeologs within wheat triads, our question was in identifying homoeologs which behaved differently when placed under free-running conditions. We therefore still feel justified in using the full 0-68h dataset to identify homoeolog periods and phases which indicate differential circadian regulation, but we have now clarified that we are using the full dataset for the triad analysis in the results (Ln 140).

Fig 1h-m. How were those genes chosen? It would help to see the SD of the replicates shown, since this is just showing one triad. It would be helpful to see a plot that represents the full set of triads rather than just one that looks best. If normalized to a standard phase they could be put on the same plot. For example, panel j is meant to show the 8h lag of subgenome D. If the data is normalized so that A and B are set to the same phase all the triads could be displayed with shaded SD bars to show the variation. Something like this would be a better representation of the data rather than showing just one example.

Fig. 1h-m are case-studies illustrating the different forms of circadian imbalance between homoeologs. We agree that it is helpful to see the standard deviation as error bars on these triad plots and have added it as suggested. In line with another Reviewer 2’s suggestion we have removed Fig 1k and have replaced this with a comparison of mean normalised data for Triad 408 and Triad 2454, highlighting the difference between imbalanced rhythmicity and imbalanced amplitudes between homoeologs. Fig 1 I and m do not have error bars as adding standard deviations to mean normalised data wasn’t appropriate.

Thank you for your suggestion on how to display the different phases between homoeologs. We feel that if we were to plot all of the triads displaying imbalanced phases, the differences in period length and accompanying noise differences would make the plot so busy as to be unreadable. We hope that the pie charts Fig 1 d-g give a global overview of the proportions of triads with circadian imbalance, but agree with the point that it is useful to allow readers to view triads of their own preference. Therefore, we have now provided the replicate level TPM data with the triad IDs annotated (Supplementary File 12) and Supplementary file 11 provides the classification of each triad alongside Metacycle statistics, ortholog identification and cluster information discussed elsewhere in the paper. Readers can now look up a triad or gene of interest and see how it was classified and what the expression looks like over the full dataset.

It is surprising that there aren't more comparisons with the B. rapa dataset, especially when discussing the clock genes that show balanced or imbalanced expression. Are they similar in B. rapa and does it support your hypothesis that unbalance for certain genes are selected against?

While we agree that a thorough, multiple species, comparative transcriptomic analysis is undoubtably of interest for the future, we feel it is beyond the scope of the questions being addressed in this paper. We do compare paralogs defined as “similar” in the Greenham dataset with homoeologs described as “balanced” in our dataset and find that genes involved with “photosynthesis” and “generation of precursor metabolites and energy” tend to be common between the two groups, potentially suggesting conservation of balance for certain types of genes (Ln 206-217).

Figure 2 networks. Why were these specific modules selected? Is it actually appropriate to directly compare these modules? I do see that some of the comparisons have high correlations from panel a, but not all. For example, in panel b the W9 and A9 modules have a correlation value of 0.92, which seems appropriate. However, panel c (modules W3 and A2) have a correlation of 0.42, which seems far too low to make any sort of comparison meaningful.

The modules were selected to simplify the comparison of genes expressed in the dawn, midday, dusk, and night. We were interested in identifying common GO-enrichment in genes peaking throughout the day, although as you have identified, the differences in period length between Arabidopsis and wheat made this difficult. Our reasons for comparing module W3 with module A2, were that, even though their eigengenes are not highly correlated per se, when period length is taken into account, both modules peak during the subjective day (CT 6.34h and 6.19h) and they share commonly enriched GO terms which make sense for day peaking genes.

Further, as described in methods comments, using a cutHeight as low as 0.15 will likely lead to some number of genes in any given module that do not necessarily "share" a similar expression pattern. These genes could have a pattern that has very low correlation to their module eigengene and were only placed in that module because the pattern was "less similar" to other module eigengenes. The current expression plots in this figure follow a clear pattern, but I suspect this would be even more apparent if the genes within these modules had a higher correlation to the module eigengene. Perhaps the current genes in these modules could just be filtered to have a higher correlation score?

Thank you for your comments, we have now made changes to the Results and Methods to clarify our approach (Ln 237-239 and Ln738-765). Merging modules with highly correlated module eigengenes (ME) is the final step in constructing our co-expression networks. To do this, as the reviewer describes - we used the WGCNA default parameter of a mergeCutHeight() of 0.15. This results in the merging of modules with highly correlated ME as the 0.15 mergeCutHeight() refers to the dissimilarity metric of 1 minus the eigengene correlation. So for WGCNA, a mergeCutHeight() of 0.15 corresponded to a correlation of 0.85. For the wheat modules, we took the additional step of merging closely related modules (mergeCloseModules()) using a cutHeight of 0.25, again a dissimilarity metric of 1 minus the eigengene correlation (corresponding to a correlation of 0.75). Reducing the stringency of the cutHeight to merge highly correlated wheat modules enabled us to more easily compare significantly correlated wheat and Arabidopsis co-expression modules to identify groups of genes in wheat and Arabidopsis expressed at similar times in the day, and enable the comparison of whether similar phased transcripts in wheat and Arabidopsis had similar biological roles.

Lines 327-334: I am not following the connection between 'response to abiotic stimulus' and the photoreceptor and light signaling proteins. At the start of this section (line 308) the authors say that the GO analysis was only done on rhythmically expressed genes but the reference to only one PHYA being rhythmic and yet multiple genes are shown in the plot in fig. S16. Does this mean that all the genes were shown and not just the rhythmic ones? This would explain why many of the PHY and CRY genes don't seem to have rhythms. This should be clarified better in the text or indicated in the plot which ones were called rhythmic. Since the first day following transfer is still the diel pattern from the entrainment condition, what does the PHY and CRY expression look like? Does it appear rhythmic under diel but lose rhythmicity in LL? It should be noted in the text that arrhythmicity in circadian conditions doesn't mean there isn't rhythmicity under diel conditions. This could be an additional explanation apart from the current one in the text that the regulation is at the level of protein stability/localization. Overall, this entire section is very long and entirely based on data shown in the supplemental material. I do appreciate having the individual gene plots that supplement Figure 4 and would suggest either providing a main figure to highlight a small subset of genes or pathways in this section or shorten it and focus on the results shown in the main figures.

Upon reading the reviewer’s comment, we realize that we should have made our motivations and processes clearer within this section. We used the data filtered for rhythmicity to conduct the GO-enrichment analysis and then used that to identify processes which should be of interest for further investigation. We have now added an additional sentence (Ln 352-354) to explain this more clearly. We then considered the orthologs of well-known Arabidopsis gene networks and extracted their expression from our circadian dataset, whether rhythmic or not. Supplementary Table 10 contains all of the genes we investigated, their expression and their MetaCycle statistics. We have also indicated here which genes are plotted in which Supplementary Figure 18-20. The reasons for plotting non-rhythmic genes in some cases was that it illustrates the differences between circadian control in Arabidopsis versus wheat (as is the case for the PHY and CRY genes). We understand that it is useful to see at a glance which genes are classified as rhythmic or arrhythmic, so have now highlighted each row in Supplementary Table 10 to make this more intuitive, and added a read me tab.

Regarding your point about oscillation under diel cycles, we agree that some transcripts will show rhythmic behaviour under entraining environments but not under constant conditions, and may perform time-of-day specific functions. However, these transcripts are likely to not be regulated by the circadian clock (at the transcriptional level) and so are not discussed in the context of a circadian transcriptome.

For your interest, here is the full expression of PHY and CRY transcripts starting at ZT0:

[Image]

It is difficult to say for definite, but it seems likely that some of these photoreceptors will have rhythmic patterns of expression under diel cycles, but these rhythms do not endogenously persist under constant conditions.

We appreciate your feedback that this section would benefit from cutting down of text and addition of a Figure to illustrate the text. We have now cut some of this section down and created a new main figure based on some of the oscillation plots from Supplementary Figure 18 and 19. We chose examples that reflect a conservation of relationships between transcripts of different peak phases, as we find it interesting that both species have similar patterns. (Main Figure 4, Ln 361--363, 382).

1. Primary metabolism section: in terms of the supplemental figure, similar to the previous one I think it would declutter the plots if the genes that are not rhythmic were left out and simply indicate below the plot that they didn't meet the rhythmicity cutoff. This is another area where there is more discussion surrounding the supplemental figures than the main figure 4.

One of the overall findings of this section was that many of the genes involved in Starch and T6P metabolism which are rhythmically expressed in Arabidopsis are not rhythmically expressed in wheat. We feel removing these genes from the results would detract from the importance of this finding. We have now edited Supplementary Table 10 to highlight which genes are classified as rhythmic. We have also added in a sentence to the start of this section which lays out our motivations for this analysis, summarises our findings and better connects the text with an explanation of Fig. 5 (Ln 408-430).

For all gene expression figures there should be SD or SE shown either as bars or ribbons to represent the variation in replicates.

Although we agree that error bars are informative for showing variation between replicates (and have added them to Fig. 1 to show differences within wheat triads) we feel that adding error bars to the gene expression plots in Fig. 3, Fig 4 and Supplementary Fig 19-20 would make these plots difficult to read, particularly where the wheat homeologs are very similar. The purpose of these gene expression plots is to compare circadian profiles in Arabidopsis and wheat orthologs rather than to claim significant differences in expression at any particular timepoint. This is fairly common in other circadian biology studies:

https://www.pnas.org/doi/10.1073/pnas.1408886111 ,

https://www.jbc.org/article/S0021-9258(17)49454-3/fulltext#seccestitle20 , https://journals.plos.org/plosone/article/comments?id=10.1371/journal.pone.0169923 , https://www.science.org/doi/10.1126/science.290.5499.2110?url_ver=Z39.88-2003&rfr_id=ori:rid:crossref.org&rfr_dat=cr_pub%20%200pubmed,

https://www.frontiersin.org/articles/10.3389/fgene.2021.664334/full,

https://www.science.org/doi/full/10.1126/science.1161403

The replication level information for each gene has now been made available in Supplementary file 12.

1. It would be very helpful to include the code used to generate the networks and perform the cross-correlation of eigengenes across networks should be included in the Methods. This will also save you from responding to email requests!

Thank you for your comment, Code for the cross-correlation analysis, Loom plots and WGCNA network construction is now available from our groups GitHub repository: https://github.com/AHallLab/circadian_transcriptome_regulation_paper_2022/tree/main

Minor Comments

1. Figure 1, panel d: - The "unbalanced" triads that are depicted by the lighter shading; do these in fact have a different cutoff than the original rhythmic homoeologs? In the figure it says qThank you for bringing this to our attention, this has now been corrected.

Hard to directly compare the GO term overlap in Figure 2f. Might be better to only show the results for the 4 pairs shown in b-e and put them side by side in the bubble plot.

Thank you for this feedback, We have tried to make this plot easier to understand without losing any of the available information. Hopefully it is now more intuitive to understand which columns are being compared. We have changed the coloured lines to make them slightly wider, put the modules in corresponding coloured boxes and highlighted GO-slim terms shared by modules being compared.

1. Line 314 -316 don't see supp tables 10, 11

Our apologies, these files were missed previously from the upload are now available.

1. For the selection of B. rapa circadian paralogs with similar and differential expression patterns (starting line 714), the authors choose a hard cut off of 0.001 (differentially patterned) OR 0.1 (similarly patterned). What happens to the genes that are between these two cut offs or is this a typo. Since all the other cutoffs for rhythmicity was set at 0.01 it seems likely that this is a typo.

We have now clarified this in the methods, (Ln 807-822). This is not a typo, but it is a different method to the Metacycle approach we have used for our wheat data. We defined similar/different paralogs as characterized in Greenham et al, (2020) using DiPALM p-values. We chose these DiPALM p-value cut-offs as they gave us approximately equal numbers of paralogs in each category, which represent tails of similarly expressed or differently expressed circadian genes. We checked these cut-offs by calculating average Pearson’s correlation statistics between paralogs and found that differential Brassica paralogs had a mean Pearson correlation coefficient of 0.31 (SD = 0.43) and similar Brassica paralogs had a mean Pearson correlation of 0.75 (SD= 0.23) which confirms that the DiPALM method of defining expression patterns makes sense in the context of this analysis.

Line 681. Should be supplemental Figure 6 not 9.

1. References to most supplemental figures are not the correct number.

2. Labels above the plots in Supp Fig5 do not match the legend.

We apologise for these mistakes. We realize that we had mistakenly submitted an earlier draft of the Supplementary materials file, which was missing Supplementary Figure 5, 6 and 9 which therefore shifted the order of the remaining figures. This is now updated.

1. Suppl table 7 should be as a separate .csv file or similar to be able to see the full table.

This is a good suggestion, and we have added this.

1. Line 723 should be B. rapa not B. napus.

Thank you for catching this! Corrected.

1. Figure 4. There is no explanation for what the black boxes represent in the figure legend.

Thank you for your comment. Figure 4 (new Figure 5) has now been updated.

Reviewer #1 (Significance (Required)):

This study provides new insight into the circadian regulation of the transcriptome in a new allopolyploid. It adds a valuable resource to a growing collection of circadian studies in important crops and will greatly improve our efforts to learn more about the circadian control of important crop improvement traits. The dataset will be of interest to other plant circadian biologists as well as the general plant biology community who focus on monocot crops. My expertise is more on the transcriptomic side and I do not have the expertise to evaluate the phylogenetic work presented in this study.

Reviewer #2 (Evidence, reproducibility and clarity (Required)):

Summary Rees et al. present an RNAseq time course of bread wheat. Its recent polyploidisation is one motivation for this study as gene expression dosage is known to be important for clock function in other plants. The time course covers 3 days at sampling intervals of 4h of 2-week old wheat plants (all aerial tissues), in triplicates. The subsequent analysis of the RNAseq data includes analysis of the generated data by itself (e.g. GO analysis, rhythmicity, period and phase analysis, rhythmicity of transcription factor families as well as TF binding sites) as well as thorough comparison with published datasets of other species (Arabidopsis, Brassica rapa, Brachypodium dystachion). One of the key findings is that the mean period length and the period spread are larger in wheat than in these other species). Circadian clock genes largely have similar dynamics in wheat compared to Arabidopsis. In addition, one focus is the analysis of the dynamics of three genes of one triad and imbalance / balance of such triads. To the surprise of the authors, circadian regulated and clock genes were not necessarily balanced. Silencing is one of their explanation for imbalance of circadian genes as arrhythmic genes of one triad are typically those with the lowest expression level. Finally, the authors point out more examples of rhythmic processes and genes (photoreceptors and signalling, auxin, carbon metabolism) and their commonalities and differences with Arabidopsis.

Major comments - The key conclusions and the data are convincing

We thank the reviewer for their supportive comments.

• line 120 and figure 1: In my opinion, q > 0.05 is not a good definition of arrhythmicity as non-significant q-values can result from either noise in spite of rhythmicity or from arrhythmicity. A more statistically sound way to detect arrhythmicity could for example be two-one-side tests (for example in the R package 'equivalence', e.g. see usage for time courses by Noordally et al. 2018, https://www.biorxiv.org/content/10.1101/287862v1).

Thank you for pointing us in the direction of this package, we agree that choosing methods for circadian quantification and q-value cut-offs is always tricky and different approaches will perform better for noisier or non-sinusoidal waveforms. For future work, we will investigate the application of the suggested method in circadian rhythmicity analysis. However, we believe that the criteria used in this paper for rhythmicity quantification is suitable for addressing our questions, and overall, we are satisfied that rhythms with a q-value of >0.05 would also be classified by eye as being arrhythmic, and rhythms with a q-value Many other studies have used meta2d B.H q-values as a metric of rhythmicity: e.g. (https://bmcplantbiol.biomedcentral.com/articles/10.1186/s12870-022-03565-1 , https://link.springer.com/content/pdf/10.1186%2Fs12915-022-01258-7 , https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8782462/pdf/pcbi.1009762.pdf )

• lines 480-484 and intro: In the introduction, the authors write that expression levels of clock components are important for the function of the clock, and that this is one motivation for the current study where polyploidisation is expected to affect the expression levels of clock genes and their outputs. I wonder what answers or speculations this study provides in the end, or whether such answers / speculations should be made clearer. For example, do the authors think that the higher variability of periods in wheat could be a consequence of lower robustness (in addition to possible spatial differences that are mentioned) due to polyploidisation? Is anything known about the period of rhythms of close wheat relatives that did not undergo polyploidisation? Did you look at dampening over the time course in wheat vs. Arabidopsis?

The point above is an interesting one, and we thank the reviewer for raising it. We agree that the high variability of periods in wheat may be a product of polyploidisation, as functional redundancy between homoeologs may allow a tolerance for less tightly regulated, non-dominantly expressed circadian transcripts. We have now added this hypothesis to our discussion: Ln536-550.

In our comparative analysis of period distributions, we looked at periods of transcripts from a diploid relative of hexaploid wheat, Brachypodium distachyon. In Brachypodium, period lengths have around the same SD as in Arabidopsis but the mean period length is slightly longer (Supplementary table 2). We have now edited our results to make the relationship between wheat and Brachypodium clearer (ln 109-110).

Minor comments:

Introduction - lines 49: it is unclear what is meant by ppd-1 at this position of the sentence

We agree this was unclear and have revised it to “notably the ppd-1 locus within TaPRR3/7” Ln 52

• line 54/55: clarify that this refers to Arabidopsis thaliana

Corrected.

Results - line 69 and 76: cite references for these tools here (not only in the methods section)

Corrected.

• line 90-93: Why wouldn't the same thing happen on subsequent subjective evenings?

Thank you for your comments. We have now combined our hypothesis for why wheat transcripts expressed at dusk have longer periods with the discussion about longer periods upon transfer to constant light. We think that the two processes are likely to be connected and have now placed them together in Ln 126-131.

The behaviour of mean period lengths of wheat transcripts upon transfer to constant light was unexpected and we believe is quite interesting. One explanation is that the influence of the ongoing light zeitgeber when dusk was expected causes a delay in the expression of evening peaking genes which are delayed by the continuous light signal. Then, on subsequent evenings the influence of the diel dusk signal is ‘forgotten’ as the governance of the endogenous clock takes over. The very long period observed at 0-24h (28.61h) may be due to a phase shift rather than an intrinsic lengthening of period per se. Whether this trait is unique to wheat or can also be seen in other plant species is, to our knowledge, unknown.

• line 118: what is your defined cutoff for significance of the Chi square test (p=0.03 not regarded significant?)

The reviewer is completely right, we have now clarified this. Ln 145-149

• figure 1h,i: In order for the reader to see whether A and D (Figure 1h) or A (figure 1i) are indeed arrhythmic, one would need to see plots with a normalisation as done in figure 1m for 1l.

We have now removed the triad showing one rhythmic gene and two arhythmic genes (as Fig. 1h already illustrates this type of circadian imbalance) and replaced this with a side by side comparison of how imbalance in rhythmicity differs from imbalance in relative amplitude as suggested.

• figure 1h-m (and others with circadian time course traces): could a measure of variation (e.g. SD, SEM, confidence interval) be plotted as a shaded region around the curves (unless they're so small that they are there but not visible)?

We have now added error bars to these plots to show standard deviation between replicates, in Fig. 1 h, j, k and l. We could not think of an accurate way to display this information for the mean normalised data (Fig 1. i and m) so have not put error bars on these plots.

• line 139 (also in 737 and 450): give reference to Ramirez-Gonzalez et al in the same style as the rest of the manuscript (number)

Thank you for raising this, we believe we have corrected all in-text citations (both narrative and fully parenthetical form) for consistency with the APA format used by the majority of Review Commons affiliate journals.

• Clustering (modules): What is the reason for choosing 9 clusters? Was this number optimised or chosen for other reasons?

WGCNA uses an unsupervised clustering algorithm that works within the supplied parameters to determine the optimum number of clusters to explain the dataset, without prior specification of the number of clusters. We have amended the manuscript text to clarify this Ln237-239.

• lines 280 - 284: The TaELF3-1D phenotype could be explained a bit better to the non-wheat specialist, for example by mentioning in the beginning of this set of sentences.

Done (Ln 314-318).

• The authors present an analysis of TF binding sites. Can they say something about binding sites in a less sophisticated manner, such as on some very well-known motifs in promoters like the evening element?

We agree that this is a very interesting question, and one that we may investigate in more detail with our data in the future. In this paper, we performed a global analysis of wheat TFBS predicted from orthologous Arabidopsis TF targets. These targets have been experimentally validated in Arabidopsis using DAP-seq, but we have not validated that these binding sites exist in wheat promoters. We therefore took a tentative approach, and presented only enrichments at the superfamily level rather than talking about specific regulatory motifs.

The evening element would fit most likely fit within the MYB or MYB-related TFBS superfamily, however the diversity of transcription factors in this family means that there is significant enrichment of these TFBS in multiple modules throughout the day (Supplementary Figure 11). In summary, a more in depth TFBS analysis of known circadian motifs is of great interest, but we feel would be a substantial work in its own right.

• Figure 1h-l: If known or meaningful, it would be interesting to know the gene identities behind the triads shown, as in supplementary figure 5.

These triads were selected as case studies to exemplify the ways in which we were defining imbalanced circadian triads. They have no particular relevance to the figure, but out of curiosity, these are the closest Arabidopsis orthologs for the triads displayed in Fig. 1:

Triad 408 has highest identity to a hypothetical protein (AT4G26415).

Triad 2454 is similar to AT3G07600, a heavy metal transport/detoxification superfamily protein

Triad 13405 is similar to AT3G22360, encoding an ALTERNATIVE OXIDASE 1B, AOX1B

Triad 10854 is similar to NSE4A, a δ-kleisin component of the SMC5/6 complex, possibly involved in synaptonemal complex formation (AT1G51130).

Information about wheat gene names in each triad and their Arabidopsis orthologs can be viewed in Supplementary Table 11, so that readers can search for genes of particular interest to them.

• Figure 4 and text: The illustration of starch metabolism is very helpful. However, I think the paper would benefit from giving a better reason for the selection of this specific set of processes, for example by relating these findings to functional differences in starch metabolism in the two species (in contrast to Arabidopsis, wheat stores little starch in leaves but uses fructans as main reserve carbohydrate)? Are there known differences in the dynamics of starch degradation during the night?

The reviewer raises an interesting point, and we have now clarified in our results that the stated differences between starch regulation in Arabidopsis and wheat was part of the motivation behind studying this pathway. Starch is at the centre of plant primary metabolism as a carbon storage source and is arguably one of the most important features that breeders look for in regard to grain filling and yields. Additionally, it is of interest to circadian biologists as starch (as well as sucrose) have been shown to transiently cycle and to be regulated by the circadian clock. However, in wheat, carbon storage primarily uses sucrose rather than starch, and we have now added sucrose to Figure 5 to place it in this context. We think your suggestion has now improved our explanation for why we focused on starch in the manuscript, and we are grateful for your input (Ln 408-421).

We also agree that the differences in the ways that Arbaidopsis and wheat utilise starch versus sucrose, and perhaps the role that fructans have in as a reserve carbohydrate and in protection against freezing in wheat may be one of the reasons we are seeing differences in circadian regulation of starch. We have now added this to our discussion (Ln 584-592).

• Figure 4: triose-phosphates can be transported in and out of the chloroplast, as is illustrated in the figure. However, the illustration looks as though they are converted to hexose phosphates during the transport process. In order to be consistent with other transport processes of the figure (maltose and glucose), triose-phosphate should be repeated on the cytosolic side.

We have now amended this (new Fig. 5). Thank you for your feedback.

Methods - line 543: if I understand correctly that triplicates were collected and analysed for each time point, '18 samples' is mis-leading (18 time points would be more accurate).

We agree this was badly worded. Changed Ln 615.

Supplementary - Supplementary figure 3: x axis label very small and contains typo

Now corrected. Also enlarged axis for Supplementary Figure 2.

• Supplementary table 1: Romanowski et al 2020 (add year), or use ref. number citation style as in the rest of the manuscript

Thank you for raising this, we have now hopefully corrected all in text citations (both narrative and fully parenthetical form) to be consistent with APA format used by the majority of Review commons affiliate journals.

• Supplementary table 9, primary metabolism: does bold highlighting of Arabidopsis accession numbers have a meaning or is it accidental?

We apologise that this was unclear. We have corrected this. Supplementary Table 10 now also has a “Read me” tab which explains that table.

Reviewer #2 (Significance (Required)):

I believe this is a precious, carefully generated and analysed dataset which many biologists will benefit from, beyond wheat or circadian specialists. The dataset expands the knowledge of circadian transcriptome regulation to an important crop and contributes a resource of which only a handful of others exist in other species. Many high impact papers on RNAseq include some follow-up on candidates, for example in Romanowski et al 2020, which is admittedly easier to do in Arabidopsis than wheat due to the availability of genetic resources.

My expertise: Plant circadian clock (Arabidopsis), dataset analysis (but not specifically for RNAseq)

Reviewer #3 (Evidence, reproducibility and clarity (Required)):

This manuscript is based on the analysis of a single experiment consisting in transcriptomic profiling of one (hexaploid) wheat genotype along 3 days (samples taken every 4 hours). The experiment is performed in constant light conditions, allowing detection of transcripts controlled by the circadian clock. The bioinformatic analysis studies the dynamics of the different homoeologous transcript in the polyploid genome and compares cycling transcripts in wheat with what is known from Arabidopsis.

The manuscript is well written, the methods are correct, the analysis performed is sufficiently extensive and the figures are clear. The manuscript finds interesting expression patterns among homeologous genes, and goes into detail on important differences in circadian regulation of relevant gene families between Arabidopsis and wheat. The work is purely descriptive and does not aim at associations with physiological phenotypes, but the bioinformatic analysis is very thorough and uncovers interesting examples.

Only one caveat: For what I gather, there is no replication in the RNA-seq experiment, although the exact method does not appear in the text. From the Methods section: "tissue was sampled every 4h for 3 days (18 samples in total)" and "At each timepoint, we sampled the entire aerial tissue from 3 replicate plants". Whether these samples were pooled or not is not described. The "Data Availability" section links to 18 RNA-seq paired end libraries, which suggest that the replicates were pooled, although some type of barcoding might have been used. The text should mention if the replicates were pooled or not, and, if so, what was the method used for poling (tissue, RNA or libraries). Even in the case of no biological replication the manuscript brings interesting insights into wheat transcriptomics and circadian biology. The editor (or the rules of the journal) should decide if they accept articles with no "real" biological replication (I am sure we all understand by now the benefits and limitations of pooling biological replicates into a single RNA-seq library).

There was replication within the RNA sequencing experiment, and we apologise that this was unclear from our manuscript. Each timepoint consisted of three independent biological replicates. We have now created a new “Experimental context” section in the results to explain this (Ln 74-82) and have clarified in the methods how our data was processed (Ln 609-615 and 636-638).

We have now included an additional matrix with TPMs at the replicate level to assist readers in looking at specific genes of interest (Supplementary Table 12).

Minor comments:

The description of the experimental setup in the first sentence of the Results section is too brief. Could you please talk about for how long the experiment was running? At what intervals the samples were taken? What conditions were used?

We apologise that this was unclear. We hope that the new Experimental Context section, added in response to comments from several reviewers, makes this much clearer, alongside the clarification in the methods (Ln 609-615 and 636-638).

Line 280: "...due *to* an introgression..."

Corrected. Ln 315

The legend of Figure 3l says elf4 instead of elf3

We thank the reviewer for noticing this mistake that we have now corrected.

Line 306 "says Supplementary Note 7 instead of Supplementary Note 7

We are not sure what is to be corrected here!

Reviewer #3 (Significance (Required)):

This works advances our knowledge on how genome wide expression levels are controlled by the circadian clock in polyploids. Although previous works had performed similar analyses in other polyploid plants, this is the first time this is done in an hexaploid. This work is a starting step to understand gene regulation in this important crop, and have interest for researchers working in fundamental and applied plant biology.

Thank you for your positive comments and your feedback in improving this manuscript. We would like to clarify that to our knowledge, this work presents the first analysis of a circadian transcriptome in a polyploid crop. The work by Greenham et al, although undoubtably providing insight into circadian regulation of ancient paralogs, was performed in the diploid Brassica rapa.

Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

Learn more at Review Commons

Referee #1

Evidence, reproducibility and clarity

In this study, Rees et al. perform an RNA-seq circadian time course experiment in the recently formed allopolyploid wheat. Through comparisons with other circadian transcriptomic datasets in other species it appears that the period of rhythmic genes is much more variable in wheat with a shift to longer periods compared to the other species examined. Interestingly, by analyzing circadian parameters among expressed genes, they find evidence that this newly formed allopolyploid already shows signs of divergence in circadian traits among homoeologs. A thorough comparison with circadian regulated genes in Arabidopsis reveals overlap in phasing of genes involved in certain biological processes such as photosynthesis and light signaling whereas genes involved in starch metabolism were found to have different levels of rhythmicity and phasing. This dataset will be a great resource for the community and enable new predictions about the influence of polyploidy on the circadian control of important crop improvement traits and the circadian regulation of gene expression.

Major Comments

1. The results section starts with very little explanation of the experiment. It would help to provide a little more detail at the start of the results to explain the context for the experiment and what was done, when samples were collected and for how long. For the methods section, it isn't until line 650 that it is clearly stated that the sampling started at ZT0. It would be better to put this in the plant materials and growth condition section.
2. The low proportion of circadian regulated genes is likely due to the very low cutoff for calling a gene expressed, especially when there are three days of repeated timepoints. If a gene is expressed across the time course it should have values above TPM 0 for at least 3 time points in order for it to be expressed each day. I'd also be suspicious of a gene with a TPM value less than 0.5. Comparing these types of numbers is always challenging due to the various cutoffs used. Along those lines, why was a different filtering scheme used for Arabidopsis (line 657)?
3. In reference to the shortening of the period every day, this should be interpreted with caution. Period estimate of a single cycle are not very reliable and the SD for each day is around 3h so it is difficult to draw any conclusions about changes in period each day. One option would be to only include genes with an SD less than 1h or alternatively to remove the discussion surrounding the comparison of period across the three days and focus on the period results for the full 24h-68h window shown in 1b. While 2 days is better it is still not ideal for calling period; however, your first day will still have a strong diurnal driven pattern that will likely skew your circadian period.
4. Line 87-93: If the dusk cue is important for clock expression you would think this would be biased towards genes that peak later in the day or near dusk. This argument should be connected better to the period results discussed on lines 98-101.
5. Lines 650-652 of the Methods mentions that one of the main interests was the response to transfer to L:L, but this isn't mentioned in the introduction and doesn't come up much in the Results section. Most of the expression comparisons are focused on the 24-68h window. It also isn't clearly explained why the first day in LL is still a diurnal cycle. This would be helpful for non-circadian readers who may wonder why the first day is not included in all the analyses.
6. The phase comparisons shown in Figure suppl 4 are confusing. Suppl. Note 3 states that the period from the 24-68h data window was used to establish the bins but then the phase is shown for 3 different windows for each column? When calculating the phase for each of those 3 windows which period was used as the denominator in the phase calculation? Was it the period that matches the window used to calculate phase? What does the plot look like if phase is called on the same window used to calculate period (24-68)? What method was used to call phase in Suppl. Fig 4? As shown in Suppl Fig. 3 the method can influence the phase distributions. The methods suggest that the phase was determined with Metacycle but then FFT and MESA were used to verify. What does this mean verify, were they adjusted if FFT/MESA didn't agree?
7. It is difficult to interpret the value of the period and phase comparisons shown in Fig. 1b, c, e and f after the preceding section about how variable the period and phase is across days. It is also surprising that the full 3 days were used to calculate the circadian statistics considering the first day is still under diurnal control. Do the ratios remain the same if the statistics are performed only on the 24h-68h window? For consistency with the rest of the paper and avoid confusion it would be best to have all circadian parameters measured using the same time window (24h-68h).
8. Fig 1h-m. How were those genes chosen? It would help to see the SD of the replicates shown, since this is just showing one triad. It would be helpful to see a plot that represents the full set of triads rather than just one that looks best. If normalized to a standard phase they could be put on the same plot. For example, panel j is meant to show the 8h lag of subgenome D. If the data is normalized so that A and B are set to the same phase all the triads could be displayed with shaded SD bars to show the variation. Something like this would be a better representation of the data rather than showing just one example.
9. It is surprising that there aren't more comparisons with the B. rapa dataset, especially when discussing the clock genes that show balanced or imbalanced expression. Are they similar in B. rapa and does it support your hypothesis that unbalance for certain genes are selected against?
10. Figure 2 networks. Why were these specific modules selected? Is it actually appropriate to directly compare these modules? I do see that some of the comparisons have high correlations from panel a, but not all. For example, in panel b the W9 and A9 modules have a correlation value of 0.92, which seems appropriate. However, panel c (modules W3 and A2) have a correlation of 0.42, which seems far too low to make any sort of comparison meaningful. Further, as described in methods comments, using a cutHeight as low as 0.15 will likely lead to some number of genes in any given module that do not necessarily "share" a similar expression pattern. These genes could have a pattern that has very low correlation to their module eigengene and were only placed in that module because the pattern was "less similar" to other module eigengenes. The current expression plots in this figure follow a clear pattern, but I suspect this would be even more apparent if the genes within these modules had a higher correlation to the module eigengene. Perhaps the current genes in these modules could just be filtered to have a higher correlation score?
11. Lines 327-334: I am not following the connection between 'response to abiotic stimulus' and the photoreceptor and light signaling proteins. At the start of this section (line 308) the authors say that the GO analysis was only done on rhythmically expressed genes but the reference to only one PHYA being rhythmic and yet multiple genes are shown in the plot in fig. S16. Does this mean that all the genes were shown and not just the rhythmic ones? This would explain why many of the PHY and CRY genes don't seem to have rhythms. This should be clarified better in the text or indicated in the plot which ones were called rhythmic. Since the first day following transfer is still the diel pattern from the entrainment condition, what does the PHY and CRY expression look like? Does it appear rhythmic under diel but lose rhythmicity in LL? It should be noted in the text that arrhythmicity in circadian conditions doesn't mean there isn't rhythmicity under diel conditions. This could be an additional explanation apart from the current one in the text that the regulation is at the level of protein stability/localization. Overall, this entire section is very long and entirely based on data shown in the supplemental material. I do appreciate having the individual gene plots that supplement figure 4 and would suggest either providing a main figure to highlight a small subset of genes or pathways in this section or shorten it and focus on the results shown in the main figures.
12. Primary metabolism section: in terms of the supplemental figure, similar to the previous one I think it would declutter the plots if the genes that are not rhythmic were left out and simply indicate below the plot that they didn't meet the rhythmicity cutoff. This is another area where there is more discussion surrounding the supplemental figures than the main figure 4.
13. For all gene expression figures there should be SD or SE shown either as bars or ribbons to represent the variation in replicates.
14. It would be very helpful to include the code used to generate the networks and perform the cross-correlation of eigengenes across networks should be included in the Methods. This will also save you from responding to email requests!

Minor Comments

1. Figure 1, panel d: - The "unbalanced" triads that are depicted by the lighter shading; do these in fact have a different cutoff than the original rhythmic homoeologs? In the figure it says q<0.1 but I thought it was q<0.01.
2. Hard to directly compare the GO term overlap in Figure 2f. Might be better to only show the results for the 4 pairs shown in b-e and put them side by side in the bubble plot.
3. Line 314 -316 don't see supp tables 10, 11
4. For the selection of B. rapa circadian paralogs with similar and differential expression patterns (starting line 714), the authors choose a hard cut off of 0.001 (differentially patterned) OR 0.1 (similarly patterned). What happens to the genes that are between these two cut offs or is this a typo. Since all the other cutoffs for rhythmicity was set at 0.01 it seems likely that this is a typo.
5. Line 681. Should be supplemental Figure 6 not 9.
6. References to most supplemental figures are not the correct number.
7. Labels above the plots in Supp Fig5 do not match the legend.
8. Suppl table 7 should be as a separate .csv file or similar to be able to see the full table.
9. Line 723 should be B. rapa not B. napus.
10. Figure 4. There is no explanation for what the black boxes represent in the figure legend.

Significance

This study provides new insight into the circadian regulation of the transcriptome in a new allopolyploid. It adds a valuable resource to a growing collection of circadian studies in important crops and will greatly improve our efforts to learn more about the circadian control of important crop improvement traits. The dataset will be of interest to other plant circadian biologists as well as the general plant biology community who focus on monocot crops. My expertise is more on the transcriptomic side and I do not have the expertise to evaluate the phylogenetic work presented in this study.

也就是DPR更大，也就是用几个像素显示一个像素，如果一个屏幕分成两块，设置不同的DPR，那DPR越多，一个像素就越大。

Author Response

Reviewer #1 (Public Review):

The authors showed that longer reverberation time prolongs inhibitory receptive fields in cortex and suggest that this helps producing sound representations that are more stable to reverberation effects. The claims is qualitatively well supported by two controls based on probe responses to the same type of white noise in two different reverberation contexts and based on receptive fields measured at different time points after the switch between two reverberation conditions. The latter gives stronger results and thus constitutes a more convincing control that the longer decay of inhibition is not an artifact of stimulus statistics. The limits of the study include the use of anesthesia and the fact that cortex shows a very broad range of dereverberation effects, actually much broader than predicted by a simple model. This result confirms that reverberation produces cortical adaptation as suggested before, and suggests as a mechanistic hypothesis that rapid plasticity of inhibition underlies this adaptation. However the paper does not address whether this adaptation occurs in cortex or in subcortical structures. The fact that an effect is observed under anesthesia suggests a subcortical origin.

We agree that it is important to consider subcortical processing levels too, as we have done previously when investigating neuronal adaptation to mean sound level and contrast. However, these and other forms of adaptation are known to be organized hierarchically and are most prominent in the auditory cortex. In particular, in ferrets, the species we use in our study, contrast adaptation is a weaker and less consistent property of neurons in the inferior colliculus than of neurons in the primary auditory cortex (Rabinowitz et al., 2013, PLOS Biol. 11:e1001710). Similar results have been obtained for stimulus-specific adaptation and prediction error signaling in other species (Parras et al. 2017, Nature Comms. 8:2148; Harpaz et al., 2021, Prog. Neurobiol. 202:102049). It therefore makes considerable sense to focus here on the primary auditory cortical areas in ferrets, where adaptation to reverberation has been demonstrated before (Mesgarani et al., 2014, PNAS 111:6792-7), in order to explore the possible basis for this effect. We agree that future work should investigate whether adaptive shifts in the inhibitory components of the receptive fields with room size are a property of the cortex only or also found in subcortical auditory areas, such as the thalamus or midbrain.

We chose to record from anesthetized ferrets in order to provide the stability required for presenting the long stimulus sequences that were essential for characterizing the effects of reverberation on the responses of cortical neurons. This strategy was adopted only because we have previously shown that contrast adaptation is indistinguishable in the primary auditory cortex of awake and anesthetized ferrets (Rabinowitz et al., 2011, Neuron 70:1178-91). Furthermore, adaptation to background noise has been shown to enhance the representation of speech in the human auditory cortex independently of the attentional focus of the listeners (Khalighinejad et al., 2019, Nature Comms. 10:2509). All the same, while there is much evidence to indicate that adaptation does not differ, at least qualitatively, with brain state, it would be interesting in future research to determine how task engagement affects the inhibitory plasticity that we observed in this study.

Reviewer #2 (Public Review):

Ivanov et al. examined how auditory representations may become invariant to reverberation. They illustrate the spectrotemporal smearing caused by reverberation and explain how dereverberation may be achieved through neural tuning properties that adapt to reverberation times. In particular, inhibitory responses are expected to be more delayed for longer reverberation times. Consistently, inhibition should occur earlier for higher frequencies where reverberation times are naturally shorter. In the manuscript, these two dependent relationships were derived not directly from acoustic signals but from estimated relationships between reverberant and anechoic signal representations after introducing some basic nonlinearity of the auditory periphery. They found consistent patterns in the tuning properties of auditory cortical neurons recorded from anesthetized ferrets. The authors conclude that auditory cortical neurons adapt to reverberation by adjusting the delay of neural inhibition in a frequency-specific manner and consistent with the goal of dereverberation.

Strengths:

This main conclusion is supported by the data. The dynamic nature of the observed changes in neural tuning properties are demonstrated mainly for naturalistic sounds presented in persistent virtual auditory spaces. The use of naturalistic sounds supports the generalization of their findings to real live scenarios. In addition, three control investigations were conducted to backup their conclusions: they investigated the build-up of the adaptation effect in a paradigm switching the reverberation time after every 8 seconds; they analyzed to which degree the observed changes in tuning properties may result from differences in the stimulus sets and unknown nonlinearities; and, most convincingly, they demonstrated after-effects on anechoic probes.

Thank you.

Weaknesses:

1) The strength of neural adaptation appears overestimated in the main body of the text. The effect sizes obtained in control conditions with physically identical stimuli (anechoic probes, Fig. 3-Supp. 3B; build-up after switching, Fig. 3-Supp. 4B-C) are considerably smaller than the ones obtained for the main analyses with physically different stimuli. In fact, the effect sizes for the control conditions are similar to those attributed to the physical differences alone (Fig. 3-Supp. 2B).

The best estimates of the magnitude of the neural adaptation in our paper come from the STRF analysis, and the potential effects of stimulus differences is estimated using our simulated neurons method. While the noise burst and room switching experiments are very valuable controls for verifying the presence of the adaptation, they may underestimate the adaptation’s magnitude because the responses to the anechoic noise burst probes may become partially unadapted during their progress, lessening the adapted effects for these sounds. Likewise, the room switching control may not capture the full magnitude of the adaptive effect because the time spans of two time windows used to assess the adaptation (i.e. L1 and L2 or S1 and S2) have limited resolution and may not be optimally matched to the timescourse of the adaptation. However, the noise burst and room switching analyses are critical controls in our study, even if the measured effects may be more subtle. Crucially, these analyses demonstrate that the reverberation adaptation can be observed even for physically identical stimuli. This confirms, in addition to our simulated neuron methods, that the effects described in our manuscript cannot be entirely due to fitting artifacts resulting from comparing neural responses to different acoustic stimuli, but rather result, at least in part, from an underlying adaptive process.

2) All but one analysis depends on so-called cochleagrams that very roughly approximate the spectrotemporal transfer characteristics of the auditory periphery. Basically, logarithmic power values of a time-frequency transformation with a linear frequency scale are grouped into logarithmically spaced frequency bins. This choice of auditory signal representation appears suboptimal in various contexts:

On the one hand, for the predictions generated from the proposed "normative model" (linear convolution kernels linking anechoic with reverberant cochleagrams), the non-linearity introduced by the cochleagrams are not necessary. The same predictions can be derived from purely acoustical analyses of the binaural room impulse responses (BRIRs). Perfect dereverberation of a binaural acoustic signal is achieved by deconvolution with the BRIR (first impulse of the BRIR may be removed before deconvolution in order to maintain the direct path). On the other hand, the estimation of neural tuning properties (denoted as spectro-temporal receptive fields, STRFs) assumes a linear relationship between the cochleagram and the firing rates of cortical neurons. However, there are well-described nonlinearities and adaptation mechanisms taking place even up to the level of the auditory nerve. Not accounting for those effects likely impedes the STRF fits and makes all subsequent analyses less reliable. I trust the small but consistent effect observed for the anechoic probes (Fig. 3-Supp. 3B) the most because it does not rely on STRF fits. Finally, the simplistic nature of the cochleagram is not able to partial out the contribution of peripheral adaptation from the adaptation observed at cortical sites.

The reviewer brings up two important issues to consider here. The first is our use of cochleagrams to model peripheral input to the auditory cortex. The second is our use of STRFs to model the receptive fields of auditory cortical neurons.

In a recent study (Rahman et al., 2020, PNAS 117:28442-51), we tested a wide range of cochlear models to examine which model provides the best preprocessing stage for predicting neural responses to natural sounds in the ferret primary auditory cortex. We found that the cochlear models used to produce cochleagrams in the current manuscript performed best, outperforming even more complicated and biologically-inspired cochlear models (e.g. Bruce et al., 2018, Hearing Research 360:40-54). This therefore determined our choice of cochlear model. However, to address the reviewer’s concern, we replicated our reverberation adaptation findings using Bruce et al.’s (2018) more complex cochlear model, and we include the results of this analysis in our revised version of the manuscript.

STRFs are widely used to model the receptive fields of neurons in the auditory system, and particularly in the primary auditory cortex. Nevertheless, the reviewer is correct to point out that these linear models of neural receptive fields are limited, and many cortical neurons show nonlinear aspects in their frequency and temporal tuning. In the present study, the use of STRFs in the normative deverberation model allowed us to produce predictions for neural tuning across reverberant conditions that could be directly tested in the STRFs of real cortical neurons. It is less clear to us how an acoustical analysis of BRIRs would translate into predicted neural firing patterns. While the simple STRF model used here provided new insights into a mechanism for reverberation adaptation in the auditory cortex, it would be interesting and valuable for future studies to test non-linear receptive field properties in this context. Future studies should also examine contributions to reverberation adaptation at other levels of the auditory system, including subcortical stations.

Network Administration

• 14 months
• 15 courses

420-E03-AB Intro to Network Administration 45 420-Z24-AB Business Communication Skills and Job Search 60 420-E15-AB Windows Clients 75 420-E35-AB PC Work Station 75 420-E46-AB Cisco I 90 420-E78-AB Network Installation & Administration: I 120 420-E68-AB Network Operating System: Unix 120 420-E76-AB Cisco II 90 420-E84-AB Groupware Systems 60 420-E88-AB Network Installation & Administration: II 120 420-EA6-AB Cisco III 90 420-EB6-AB Cisco IV 90 420-EB4-AB Linux Server 60 420-EE4-AB Network Security 60 420-EEF-AB Work Term 345

I tend to take a much narrower view of the use and function of a zettelkasten for the reuse of atomic ideas. As a result, from experience I'd recommend these sorts of details are probably better suited for future search in your blog, a personal wiki, or even a commonplace book format than for use in your zettelkasten. I've outlined some of the broad idea for this in an article: Zettelkasten Overreach. On the other hand, if an outline form of these things is imminently abstractable for future very active reuse in other programming environments, then perhaps it's worthwhile, but then you'd need to reach the appropriate level of abstraction for this reuse and you may have lost the more specific details for direct recreation needed as reminders for your future self.

SciScore for 10.1101/2022.05.09.22274860: (What is this?)

Please note, not all rigor criteria are appropriate for all manuscripts.

Table 1: Rigor

Table 2: Resources

Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).

Results from LimitationRecognizer: An explicit section about the limitations of the techniques employed in this study was not found. We encourage authors to address study limitations.

Results from TrialIdentifier: No clinical trial numbers were referenced.

Results from Barzooka: We did not find any issues relating to the usage of bar graphs.

Results from JetFighter: We did not find any issues relating to colormaps.

Results from rtransparent:

• Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
• Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
• No protocol registration statement was detected.

Results from scite Reference Check: We found no unreliable references.

About SciScore

SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

SciScore for 10.1101/2022.05.08.22274803: (What is this?)

Please note, not all rigor criteria are appropriate for all manuscripts.

Table 1: Rigor

Table 2: Resources

Results from OddPub: We did not detect open data. We also did not detect open code. Researchers are encouraged to share open data when possible (see Nature blog).

Results from LimitationRecognizer: We detected the following sentences addressing limitations in the study:

This study has several limitations. First, among the limitations we highlight those inherent to the type of study carried out (cross section descriptive observational study). We did not study causality, we only limited ourselves to describing our population. Second, the nature of the database did not allow more detailed information to be obtained, such as ventilatory monitoring of days after baseline or more specific laboratory data taken on days other than those officially designated for the study (for example, on weekends, no data was recorded in this regard). The number of cases is small, so there may be independent determinants of mortality that could not be identified. The sustained work burden on health personnel by COVID-19 could also have contributed to lack of some important information on medical records on the specific days designated for obtain it. Thus, further studies should allocate dedicated resource to tackle these limitations and assure a complete data set.

Results from TrialIdentifier: No clinical trial numbers were referenced.

Results from Barzooka: We did not find any issues relating to the usage of bar graphs.

Results from JetFighter: We did not find any issues relating to colormaps.

Results from rtransparent:

• Thank you for including a conflict of interest statement. Authors are encouraged to include this statement when submitting to a journal.
• Thank you for including a funding statement. Authors are encouraged to include this statement when submitting to a journal.
• No protocol registration statement was detected.

Results from scite Reference Check: We found no unreliable references.

About SciScore

SciScore is an automated tool that is designed to assist expert reviewers by finding and presenting formulaic information scattered throughout a paper in a standard, easy to digest format. SciScore checks for the presence and correctness of RRIDs (research resource identifiers), and for rigor criteria such as sex and investigator blinding. For details on the theoretical underpinning of rigor criteria and the tools shown here, including references cited, please follow this link.

Reviewer #3 (Public Review):

The contribution provides approaches to understanding group behaviour using drumming as a case of collective dynamics. The experimental design is interestingly complemented with the novel application of several methods established in different disciplines. The key strengths of the contribution seem to be concentrated in 1) the combination of theoretical and methodological elements brought from the application of methods from neurosciences and psychology and 2) the methodological diversity and creative debate brought to the study of musical performance, including here the object of study, which looks at group drumming as a cultural trait in many societies.

Even though the experimental design and object of study do not represent an original approach, the proposed procedures and the analytical approaches shed light on elements poorly addressed in music studies. The performers' relationships, feedbacks, differences between solo and ensemble performance and interpersonal organization convey novel ideas to the field and most probably new insights to the methodological part.<br /> It must be mentioned that the authors accepted the challenge of leaving the nauseatic no-frills dyadic tests and tapping experiments in the direction of more culturally comprehensive (and complex) setups. This represents a very important strength of the paper and greatly improves the communication with performers and music studies, which have been affected by the poor impact of predictable non-musical experimental tasks (that can easily generate statistical significant measurements). More specifically, the originality of the experiment-analysis approach provided a novel framework to observe how the axis from individual to collective unfolds in interaction patterns. In special, the emergence of mutual prediction in large groups is quite interesting, although similar results might be found elsewhere.

On another side, important issues regarding the literature review, experimental design and assumptions should be addressed.<br /> I miss an important part of the literature that reports similar experiments under the thematic framework of musical expressivity/expression, groove, microtiming and timing studies. From the participatory discrepancies proposed in 1980's Keil (1987) to the work of Benadon et al (2018), Guy Madison, colleagues and others, this literature presents formidable studies that could help understand how timing and interactions are structured and conceptualized in the music studies and by musicians and experts. (I declare that I have no recent collaborations with the authors I mentioned throughout the text and that I don't feel comfortable suggesting my own contributions to the field). This is important because there are important ontological concerns in applying methods from sciences to cultural performances. One ontological issue that different cultural phenomena differ from, for example, animal behaviour. For example, the authors consider timing and synchrony in a way that does not comply with cultural concepts: p.4 "Here we consider a musical task in which timing consistency and synchrony is crucial". A large part of the literature mentioned above and evidence found in ethnographic literature indicate that the ability to modulate timing and synchrony-asynchrony elements are part of explicit cultural processes of meaning formation (see, for example, Lucas, Glaura and Clayton, Martin and Leante, Laura (2011) 'Inter-group entrainment in Afro-Brazilian Congado ritual.', Empirical musicology review., 6 (2). pp. 75-102.). Without these idiosyncrasies, what you listen to can't be considered a musical task in context and lacks basic expressivity elements that represent musical meaning on different levels (see, for example, the Swanwick's work about layers/levels of musical discourse formation). Such plain ideas about the ontology of musical activities (e.g. that musical practice is oriented by precision or synchrony) generate superficial constructs such as precision priority, dance synchrony, imaginary internal oscillators, strict predictive motor planning that are not present in cultural reports, excepting some cultures of classical European music based on notation and shaped by industrial models. The lack of proper cultural framing of the drumming task might also have induced the authors to instruct the participants to minimize "temporal variability" (musical timing) and maintain the rate of the stimulus (musical tempo), even though these limiting tasks mostly take part of musical training in some societies (examples of social drumming in non-western societies barely represent isochronous tempo or timing in any linguistic or conceptual way). The authors should examine how this instruction impacts the validity of results that describe the variability since it was affected by imposed conditions and might have limited the observed behaviour. The reporting of the results in the graphs must also allow the diagnosis of the effect of timing in such small time frame windows of action.

remote desktop

111

x

virtual desktops

it

virtual workspaces

If what we're discussing here is the decision to no longer opt in to playing along with the "Western" regime for IP, then why would they limit themselves to Linux and Wine—two products of attempts to play by the rules of the now-deprioritized regime? Why wouldn't they react by shamelessly embracing "pirated" forms of the (Windows) systems that they clearly have a revealed preference for? If hackability is the issue*, then that's ameliorated by the fact that NT/2000 source code and XP source code was leaked awhile ago—again: the only thing stopping anyone from embracing those before was a willingness to play along and recognize that some game states are unreachable when (artificially) restricting one's own options to things that are considered legal moves. But that's not important anymore, right?

* i.e. malleability, and it's not obvious that it should be—it wasn't already, so what does this change?

-In UTF8 are 3 bytes - In W1252 a byte= a char

Chris Aldrich been here too

About Docker Engine and Docker Desktop

相当于[0,2)，[2,4)

Microsoft Azure is a private and public cloud platform that helps developers and IT administrators to build deploy and manage their applications.

Azure uses a technology known as Virtualization. Virtualization separates the tight coupling between a computer's hardware and its operating system using an abstraction layer called a Hypervisor. The Hypervisor emulates all the functions of a real computer and CPU and a virtual machine, optimizing the capacity of the abstracted Hardware. It can run multiple virtual machines at the same time and each virtual machine can run any compatible operating system, Such as Windows or Linux. Azure takes this virtualization technology and repeats it on a massive scale in Microsoft data centers throughout the world.

Each data center has many racks filled with servers and each server includes a Hypervisor to run multiple virtual machines. A network switch provides connectivity to all those servers. One server in each rack runs a special piece of software called a Fabric Controller. Each Fabric Controller is connected to another special piece of software known as the Orchestrator.

The Orchestrator is responsible for managing everything that happens in Azure including responding to user requests.

Users make requests using the Orchestrator's web API. The web API can be called by many tools, including the user interface of the Azure portal.

So, when a user makes a request to create a virtual machine, the Orchestrator packages everything that's needed, picks the best server rack, and then sends the packaging request to the Fabric Controller. Once the Fabric Controller has created the virtual machine, the user can connect to it. Azure makes it easy for developers and it administrators to be agile when they build deploy and manage their applications and services.

In fact, building a virtual machine is just the beginning of Azure's, ever-expanding, set of cloud services that will help you meet your business challenges. It gives you the freedom to build deploy and manage applications on a massive global network using your favorite tools and frameworks.