An understandable concern/desire: compatibility

Added new tag for this: allowing full syntax to be used, not just subset

O(x)

O(x^3)

O(log(x))

O(x)

O(x^2)

This statement assumes that different approaches to ethics / responsible use are directly related to the technical differences / nature between code and ML artefacts. I am not sure this is true.

Deze tabel lijkt er dubbel in te staan.

Start using this. See if they've got a Visual Studio Code extension.

This is extremely important because it save a lot of redundancy. We want to keep things as clear as possible. The benefit is that it increases efficiency and reduces the work of the computer.

use should in naming your tests, it makes you think about the behaviour of what you are implementing, an NLP technique according to Continuous Delivery guy

Author Response

Reviewer #1 (Public Review):

Kohler and Murray present high-throughput image-based measurements of how low-copy F plasmids move (segregate) inside E. coli cell. This active segregation ensures that each daughter cell inherit equal share of the plasmids. Previous work by different labs has shown that faithful F-plasmid segregation (as well as segregation of many other low-copy plasmids, segregation of chromosomes in many bacterial species and segregation of come supramolecular complexes) require ParA and ParB proteins (or proteins similar to them) and is achieved by an active transport mechanism. ParB is known to bind to the cargo (plasmid) and ParA forms a dimer upon ATP binding that binds to DNA (chromosome) non-specifically and also can bind to ParB (associated with cargo). After ATP hydrolysis (stimulated by the interaction with ParB), ParA dimer dissociates to monomers and from ParB and the chromosome. While different mechanisms of the ParA-dependent active transport had been proposed, recently two mechanisms become most popular - one based on the elastic dynamics of the chromatin (Lim et al. eLife 2014, Surovtsev PNAS 2016, Hu et al Biophys.J 2017, Schumaher Dev.Cell 2017) and the other based on a theoretically-derived "chemophoretic" force (Sugawara & Kaneko Biophysics 2011, Walter et al. Phys.Rev.Lett. 2017).

It is a minor comment, but we would like to point out that we do not consider these two model types as alternatives but rather as models with different levels of coarse-graining. Our interest is in the molecular-level (stochastic) models (Lim et al. eLife 2014, Surovtsev PNAS 2016, Hu et al PNAS 2015, Hu et al Biophys.J 2017, Schumacher Dev.Cell 2017).

The authors start by following motion of F plasmid with one or two plasmids per cell and by analyzing plasmid spatial distribution, plasmid displacement (referred to as velocity) as a function of their relative position, and autocorrelations of the position and the displacement. They concluded that these metrics are consistent with 'true positioning' (i.e. average displacement is biased toward the target position - center for one plasmid and 1/4 and 3/4 positions for two plasmids ) but not with 'approximate positioning' (i.e. when plasmid moves around target position, for example, in near-oscillatory fashion). This 'true positioning' can be described as a particle moving on the over-dampened spring. They reproduce this behavior by expanding the previous model for 'DNA-relay' mechanism (Lim et al. eLife 2014, Surovtsev PNAS 2016), in which plasmid is actively moved by the elastic force from the chromosome and ParA serves to transmit this force from the chromosome to the plasmid. Now, the authors explicitly consider in the model that the chromosome-bound ParA can diffuse (which the authors refer as 'hopping') and this allows the model to achieve 'true plasmid positioning' for some combination of model parameters in addition to oscillatory dynamics reported in the original paper (Surovtsev PNAS 2016).

Based on their computational model, the authors proposed that two parameters, diffusion scale of ParA = 2(2Dh/kd)1/2/L (typical length diffused by ParA before dissociation) and ratio of ParB-dependent and independent hydrolysis rates = kh/kd are key control parameters defining what qualitative behavior is observed - random diffusion, near-oscillatory behavior, or overdamped spring ('true positioning'). They vary this two parameters ~30- fold and ~200-fold range by changing Dh and kh respectively, to illustrate how dynamics of the system changes between these 3 modes of motion. While these parameters clearly play important role, the drawback is that the authors did not put either theoretical reasoning why these parameters are truly governing or showed it by varying other model parameters (kh, number of ParA NParA, spring constant of chromosome k, diffusion coefficient of the plasmid Dp) to show that only these combinations define the type of the system behavior. The authors qualitative analysis on importance of relies on the steady state solution for the diffusion equation for ParA. It is really unfortunate that no ParA distribution was measured simultaneously with the plasmid motion, as this would allow to compare experimental ParA profiles to expected quasi-steady-state solutions.

We spend almost an entire section and a figure explaining the theoretical reasoning behind the identification of the $\lambda=s/(L/2n)$ as an important system parameter (section “Hopping of ParA-ATP on the nucleoid as an explanation of regular positioning” and Figure 2) and predicted that regular positioning could only occur for $\lambda>1$. This was confirmed by parameter sweeps for the cases of 1 (Figure 3I) and multiple plasmids (Figure 5-figure supplement 1), indicating that $\lambda$ is indeed an important system parameter and that our conceptual understanding of this aspect of the system is correct. This point has now been made clearer.

However, we agree that the reasoning for $\epsilon$ (varied through the hydrolysis rate $k_h$) was not clear. It was chosen to allow us to modulate the ParA concentration at the plasmid compared to elsewhere, motivated by the differences between different ParABS systems. We originally had also considered a third quantity related to the number of nucleoid-bound ParA but we found that this had little effect on the nature of the dynamics. All three quantities describe how the timescale of a reaction/process (ParA hopping/diffusion across the nucleoid, ParB induced hydrolsysis, ParA association to the nucleoid) compares to the timescale of basal hydrolysis, which we use as a reference timescale.

We have now made this clearer as well as adding supplementary figures showing the effect of varying other system parameters at several locations in the phase diagram (Figure 3-figure supplement 3 and 4). These sweeps justify our identification of $\epsilon$ and $\lambda$ as a useful/important set of quantities for determining the dynamics of the system.

Additionally, we now add example kymographs showing the ParA distribution (Figure 3-figure supplement 2C).

The authors also show by simulations that overdamped spring dynamics can transition into oscillatory behavior when decreases, for example by cell growth. Indeed, they observed more oscillatory behavior when they compared single-plasmid dynamics in the longer cells compared to the shorter cells. This was not the case in double-plasmid cells, in eprfect agreement with their analysis. They also calculated ATP consumption in the model and concluded that the system operates close but below (perhaps, "above" should be used as it refers to bigger ) the threshold to oscillatory regime which minimize ATP consumption. While ATP consumption analysis is very intriguing, this statement (Abstract Ln24-25) seems at odds with the authors own analysis that another ParA-dependent plasmid system, pB171, operates mostly in oscillatory regime, and it is actually for this regime the authors' analysis suggest minimal ATP-consumption (Fig. 8).

To clarify, we found that pB171 (which in our hands has a copy number of 2-3 in the SR1 reduced-copy-number strain) is only clearly oscillatory in cells with a single plasmid (and only mildly so in cells with two plasmids). Otherwise, it behaves very similarly to F plasmid. We therefore believe that these two distantly related ParABS systems exhibit, overall, similar dynamics and differ only in how close the systems are to the threshold of oscillatory instability. This was not clear as we did not specify the copy number of pB171. We now provide this in Figure 7–figure supplement 1.

We refer to these systems as lying just below, rather than above, the threshold of the oscillatory instability because, on average, plasmids do not oscillate but only do so in cells with the lowest plasmid concentration.

I think the real strength of the paper is that it can potentially to show that if one considers that the intracellular cargo can be moved by the fluctuating chromosome via ParA-mediated attachments, then various dynamics can be achieved depending on combinations of several control parameters (plasmid diffusion coefficient, ParA diffusion coefficient, rate of hydrolysis and so on) including previously reported 'oscillations' (Surovtsev PNAS 2016), 'local excursions' (Hu et al Biophys.J 2017) and 'true positioning' (Schumaher Dev.Cell 2017). The main drawback (in this reviewer opinion) that this is obscured by the current presentation and discussion of this work and previous modelling work on ParA-dependent systems. For example, instead of using "unifying" potential of the presented model, yet another name 'relay and hopping' is used in addition to previously used 'DNA-relay', 'Brownian ratchet', 'Flux-based positioning', …

In the abstract and discussion, we already refer to developing a “unified” model (p1 L21, p15 L22 of the original manuscript) and in the discussion we explain how our model contains other models as limiting cases. But we agree with this recommendation - the unifying nature of our model is its main strength. We now emphasise this more.

Regarding the model name, we felt obliged to refer to the previous named models (DNA-relay and Brownian ratchet) and simply gave our model a name to avoid confusion when making comparisons. We have now removed almost all mention of ‘hopping and relay’ and just refer to ‘our model’. However, our gitlab repository with the code must have a name and therefore is still called ‘Hopping and relay’ and so the same term is used in Table 3.

… and it appears that the presented model is an alternative to these previously published work. And only in model description (in Methods section) one can find that the "... model is an extension of the previous DNA-relay model (Surovtsev et al., 2016a) that incorporates hopping and basal hydrolysis of ParA and uses analytic expressions for the fluctuations rather than a second order approximation"(p.17, ln15-17).

We are sorry that this reviewer felt that the fact that our model is an extension of DNA relay is hidden in the methods. However, we wrote in the main text:

“Motivated by the previous discussion, we decided to develop our own minimal molecular model (‘hopping and relay’) of ParABS positioning, taking the DNA relay model as a starting point … The original scheme is as follows… We supplemented this scheme with two additional components: diffusion (hopping) of DNA-bound ParA-ATP dimers across the nucleoid (with diffusion coefficient Dh, where the subscript indicates diffusion of the home position) and plasmid-independent ATP hydrolysis and dissociation (with rate kd). See Material and Methods for further details of the model. “

We now make this clearer.

However, we would argue that as models of the same system, there are naturally overlaps and the models of Hu et al and Schumacher et al could also be thought of as extensions of the DNA relay model.

While it is of course the authors right to decide how to name their model, it should be explicitly clear to the reader what is a real conceptual difference between presented and previous models from the abstract, introduction and discussion section of the paper, not from the "fine-print" details in the supplementary materials.

The main conceptual difference is that we have identified the importance of having a finite diffusive length scale for ParA diffusion/hopping on the nucleoid. This allows both oscillations and regular positioning to occur for biologically relevant parameter values and reproduces the length dependent transition from mid-cell positioning to confined oscillations that we observe for F plasmid. The DNA relay model does not have this behaviour as the ParA diffusive length scale in zero while it is infinite in the models of Ietswaart et al 2014 and Schumacher et al 2017. The model of Hu et al 2017 does have a finite length scale but the authors appear not to have realised its importance and never discovered the regular positioning regime at \lambda >1. While we make these points in the discussion in the context of Figure 8A, where we compare our model to the others, we agree with this reviewer that we should have been more explicit in the abstract and introduction. We have now corrected this.

This would allow to avoid unnecessary confusion (especially for the readers not directly involved into the modelling of ParA/B system) and clarify that all these models rely on the elastic behavior of fluctuating chromosome to drive active transport of the cargo. This reviewer believes that more explicit discussion on the models (one from the authors and previously published) differences and similarities will help with our understanding of how ParA-dependent system operate. This discussion should also include works on PomXYZ system, in which it was shown that similar dynamic system can lead to specific positioning within the cell (Schumaher Dev.Cell 2017, Kober et al. Biophys.J 2019). This will may it explicit that the models results have direct impact beyond the ParA-dependent plasmid segregation.

To further clarify the differences between the models (beyond the second and third sections of the main text and the discussion), we have now added a section to the methods and a new table (Table 3). We have also included the mentioned PomXYZ model. However, we would like this was not the first stochastic model to have ‘true’ positioning as this reviewer cites above. Though they did not include the mechanism of force generation, the model of Ietswaart et al 2014 produces regularly positioned plasmids and is referenced repeatedly in Schumacher et al. 2017.

I think that expanded parameter analysis, and explicit model comparison/discussion will make the contribution of this work to the field more clear and with the potential to advance our general understanding of how the same underlying mechanism can lead to various modes of intracellular dynamics and patterning depending on parameters combination.

Reviewer #2 (Public Review):

The work presented in this manuscript details an analysis of the partitioning of low copy plasmids under the control of the ParABS system in bacteria. Using a high throughput imaging set up they were able to track the dynamics of the partition complex of one to a few plasmids over many cell cycles. The work provides an impressive amount of quantitative data for this chemo-mechanical system. Using this data, the paper sought to clarify whether the dynamics of plasmids is due to regular positioning or noisy oscillations around a mean position. They supplement their experimental work with an intuitive model that combines elements of previous modelling efforts. Their model relies on diffusion of the ParA substrate on the nucleoid with the dynamics of the ParB partition complex being driven by the underlying elastic force due to the nucleoid on which the substrate is tethered. Their model dynamics depend on two parameters, the ratio of the length over which the substrate can explore to the characteristic length of the space and the ratio of stimulated to non-stimulated hydrolysis rates of the substrate. If the length ratio is large, ParA can fully explore the space before interacting with the ParB complex leading to balanced fluxes and regular positioning. If it gets reduced, for example by lengthening the cell, oscillations can emerge as fluxes of substrates become imbalanced and a net force can pull the partition complex.

Strengths:

Given the large amount of data, the observations unambiguously show that one particular ParABS system under the conditions studied is carrying out regular positioning of plasmids. The model synthesizes prior work into a nice intuitive picture. These model parameters can be fit to the data leading to estimates of molecular kinetic parameters that are reasonable and in line with other observations. Lining up the experimental observations with the phase space of the model suggests that the system is poised on the edge of oscillations, allowing for the system to have regular positioning with low resource consumption.

Weaknesses:

However, despite the correspondence of the simulated results with the experimental findings, other explanations are not completely ruled out. The paper emphasizes that ParA diffusion/hopping on the nucleoid is essential for the establishment of regular positioning and that without it, only oscillations were possible. Prior simulation efforts, that the paper cites, which include ParA diffusion and mixing in the cytosol but no diffusion on the nucleoid have shown that regular positioning is possible and that oscillations could get triggered as the system lengthened. Thus ParA hopping is not a necessity for regular positioning (as claimed in the paper), but very well might be needed for the given kinetic parameters of the system studied here.

We now comment on this result. In short, we believe that the mentioned model/regime is not relevant due to stochastic effects. We are not able to produce, with biological relevant parameters, regular positioning without ParA hopping.

The paper also presents experimental results for a second ParABS system (pB171) that is more likely to show oscillations. They attribute the greater likelihood of oscillations for pB1717 being due to ParA exploring a smaller space than the F plasmid system that showed regular positioning. This is pure conjecture and the paper does not provide any evidence that this is the reason. Thus it is hard to conclude if oscillations may not be due to other factors.

We do not explicitly make that claim. We did have a point in the phase diagram of Figure 8A representing pB171 with a lower value of lambda than F plasmid and stated “The location of pB171 is an estimate based on a qualitative comparison of its dynamics”. We agree this was unclear.

We now indicate the region that has oscillations with roughly the same period as single plasmids of pB171. We also make it clear that we speculate, but have not shown, that the length scale of ParA hopping is smaller than for F plasmid.

An important point here is that we can explain both oscillations and regular positioning in the same model with the same kinetic parameters, the regimes being determined by the cell length and plasmid number in a manner consistent with experimental observations.

Peer review report

Title: Crossref as a source of open bibliographic metadata

version: 2

Referee: Simon Porter

Institution: Digital Science

email: s.porter@digital-science.com

ORCID iD: https://orcid.org/0000-0002-6151-8423

General assessment

This is a clear paper that outlines a motivation (assess the metadata completeness of the Crossref record for the purposes of scientometric analysis,) along with providing a set of useful metrics to assess the completeness of each metadata field.

Essential revisions that are required to verify the manuscript

No essential revisions identified

Other suggestions to improve the manuscript

Minor suggestions: Figures in the interactive version of the preprint do not have headings or captions, or a link back to the paper.

On data availability, In the context of the paper, making the code used to process the Crossref’s XML Metadata Plus Snapshot would be a useful contribution enabling scientometric analysis of the Crossref dataset.

The following are offered as suggestions that could be added to the paper at the authors discression, but do not effect the content or the conclusions of the peer review

The authors have chosen to frame metadata completeness of Crossref records as a ‘good in itself,’ leaning on Waltman, L. (2020b) to do the work of setting this up.

Within this framework, the analysis is offered as a set of observations to help publishers understand where they need to do better. It might be the case that Publishers do not intrinsically understand why making certain metadata types available is valuable to the community.

On the question of how Crossref can be used in scientometric analysis, readers are left to make up their own minds on what Crossref can be used for today, vs what it might be capable of providing in the future based on the evidence presented. It would be a stronger conclusion to highlight the types of scientometric analysis that are now possible with Crossref, (for instance bibliometric coupling,) and those that require limits or caveats (analysis by affiliation, abstract.) As this analysis lends itself to being rerun in the future, it would be useful to trace advances (hopefully!) not just in terms of the number of things, but also in terms of how sceintometric analysis capability is progressing because of it.

Decision

Verified manuscript: The content is scientifically sound, only minor amendments (if any) are suggested.

Each password acts as a key, each character is converted into a "sudo" character that acts as a dummy key for the computer. The sudo code is verified and grants the user permission to access the app or platform if it matches up with the code stored in the computer's database.

Reviewer #2 (Public Review):

Does our proprioceptive system try to recognize our own actions?

Proprioception is our sense of the motion and posture of our own body. This sixth sense uses signals from receptors in the joints, tendons, muscles, and skin that measure forces and degrees of extension. These receptors enable us to sense, for example, the posture of our body as we wake from sleep. They also provide feedback signals that help us precisely control our limbs, for example during handwriting.

Feedback is thought to be essential to motor control, enabling the controller in our brains to rapidly adapt to the unexpected. The unexpected may include changes in the environment (like something pushing our hand that we didn't see coming), changes in our bodies (such as muscle fatigue or injury), and shortcomings of the motor program (such as a lack of precision or a badly planned limb trajectory). Feedback can come from vision and even audition, but proprioception provides an essential additional feedback path that informs us directly about the motion and posture of our limbs, and any forces on them.

How does feedback control work in the human motor system? I want to write a 'k', but there are forces on my limbs resulting from the friction of chalk on this particular blackboard. Also, my muscles are recovering from tennis practice this morning, and I haven't used chalk on a blackboard in years.

If the goal is to write a 'k', I have some flexibility. I am committed, not to a precise trajectory, but to a more abstractly defined objective: to write a legible 'k'. This suggests that feedback processing should evaluate to what extent I am succeeding at the action, not at tracing out a particular trajectory. Does what I'm actually doing look like writing a 'k'?

In a new paper, Sandbrink et al. (pp2022) report on simulations of the human musculoskeletal system and neural network models that suggest that the tuning properties of neurons in the somatosensory cortex (S1) can be explained by assuming that the objective of the proprioceptive system is to recognize the action being performed.

They used recorded traces of a person writing lower-case letters to simulate the responses of muscle spindles sensing the lengths and velocities of muscles in the human arm as would be present if the hand was moved passively along these trajectories. The physical simulation uses a 3D model of the human arm with two parameters for the direction of the upper arm and two more for the direction of the lower arm. These four parameters are inferred by inverse kinematics from the hand trajectories tracing each letter in a variety of vertical and horizontal planes. A 3D muscle model then enables the authors to compute the expected spindle responses that reflect the lengths and velocities of 25 relevant upper arm muscles.

The authors then trained neural network models of proprioceptive processing that took the simulated muscle spindle signals as input. The neural net architectures included one that first integrates information over the muscle spindles and then across time ("spatial-temporal"), one that integrated across muscle spindles and time simultaneously ("spatiotemporal") and a recurrent long-short-term-memory model.<br /> Each architecture was trained on two objectives: to decode the trajectory (i.e. the position of the hand tracing a letter as a function of time) or to recognize the action (i.e. the letter being traced). The two objectives correspond to two hypotheses about the function of proprioceptive processing: To inform the feedback controller about either the current position of the hand or the letter being drawn.

The models trained to recognize the action developed tuning more consistent with what is known about the tuning of neurons in the primary somatosensory cortex in primates. In particular, direction tuning with roughly equal numbers of units preferring each direction emerged in the middle layers of the neural network models trained to recognize the action, similar to what has been observed in primate neural recordings. Direction tuning is already present in the muscle-spindle signals, but the spindle signals do not uniformly represent the directions.

The task-optimization approach to neural network modeling is inspired by work in vision, where neural networks trained on the task of image classification explained responses to novel images in populations of neurons in the inferior temporal cortex. This result suggested a tentative answer to the why question: Why do inferior temporal neurons exhibit the response profiles and representational geometry they exhibit? Because their function (or one of their functions) is to recognize the objects in the images. Here, similarly, the authors address a why question with task-optimized neural network models: Why do somatosensory cortical neurons exhibit the types of tuning that have been reported in the literature?

The function of proprioception, of course, is not for the brain to recognize which letter it is trying to write. It already knows that. The function is to sense how the current trajectory - the actual, not the intended one - differs from, say, a legible "k" (if that was the intention), and to map from that difference to a modification vector that will improve the outcome.<br /> Why is action decoding relevant for performing the action? A key reason may be that the goal is not to produce a fixed trajectory, but to produce a legible 'k'. A legible 'k' is not a single trajectory, but a class of trajectories containing an infinity of viable solutions. If someone nudged my arm while writing, adaptive feedback control should not attempt to return me to the originally intended trajectory, but to a new trajectory that traces the most legible 'k' that is still in the cards, which may be a different style of 'k' than I originally intended.

The paper contributes a useful data set for training models and a qualitative comparison of models to real neurons in terms of tuning properties. It would be good, in follow-up studies, to directly test to what extent each of the models can quantitatively predict either single-neuron responses or population representational geometries, as has been done in vision, and to perform statistical comparisons between models.

Importantly, this paper develops the idea of combining simulations body and brain, of the musculoskeletal system, and the processing of control-related signals in the nervous system, which provides a very exciting direction for future research.

Strengths

• The paper introduces a highly original research program that marries simulation of the musculoskeletal system and neural network modelling to predict neural representations in the proprioceptive pathway.<br /> • The authors performed an architecture search and trained multiple instances of different neural network architectures with each of the two objectives.<br /> • The paper includes comprehensive analyses of the proprioceptive representations from the simulated muscle-spindle signals through the layers of the models. These analyses characterize unit tuning, linear decodability, and representational similarity.<br /> • The results suggest an explanation for the direction tuning with a roughly uniform distribution of the units' direction preferences that has been reported previously for neurons in the primate primary somatosensory (S1) cortex.<br /> • If the simulated muscle-spindle data set, models and analysis code were shared along with the published paper, this work could form the basis for quantitative model evaluation and further model development.

Weaknesses

• The models are qualitatively evaluated by comparison of model unit tuning to what is known about the tuning of neurons in the somatosensory cortex. Follow-up studies should quantitatively evaluate the models by inferential analyses of their ability to predict measured responses.<br /> • The two training objectives differ in multiple respects, making it difficult to assess what the necessary requirements are for the emergence of representations similar to primate S1. Decoding the hand position may be too simple, but what about decoding velocity, or trajectory descriptors such as curvature? There may be a middle ground between trajectory decoding and action recognition that also leads to the emergence of tuning properties as found in primate S1.

Reviewer #2 (Public Review):

Zylbertal and Bianco propose a new model of trial-to-trial neuronal variability that incorporates the spatial distance between neurons. The 7-parameter model is attractive because of its simplicity: A neuron's activity is a function of stimulus drive, neighboring neurons, and global inhibition. A neuroscientist studying almost any brain area in any model organism could make use of this model, provided that they have access to 1) simultaneously-recorded neurons and 2) the spatial locations of those neurons. I could foresee this model being the de-facto model to compare to all future models, as it is easy to code up and interpret. The paper explores the effectiveness of this distance model by modeling neural activity in the zebrafish optic tectum. They find that this distance-based model can capture 1) bursting found in spontaneous activity, 2) ongoing co-fluctuations during stimulus-evoked activity, and 3) adaptation effects during prey-catching behavior.

Strengths:

The main strength of the paper is the interpretability of the distance-based model. This model is agnostic to the brain area from which the population of neurons is recorded, making the model broadly applicable to many neuroscientists. I would certainly use this model for any baseline comparisons of trial-to-trial variability.

The model is assessed in three different contexts, including spontaneous activity and behavior. That the model provides some prediction in all three contexts is a strong indicator that this model will be useful in other contexts, including other model organisms. The model could reasonably be extended to other cognitive states (e.g., spatial attention) or accounting for other neuron properties (such as feature tuning, as mentioned in the manuscript).

The analyses and intuition to show how the distance-based model explains adaptation were insightful and concise.

Weaknesses:

Model evaluation and comparison: The paper does not fully evaluate the model or its assumptions; here, I note details in which evaluation is needed. A key assumption of the model - that correlations fall off in a gaussian manner (Fig. 1C-E - is not supported by Fig. 1C, which appears to have an exponential fall-off. Functions other than gaussian may provide better fits. Furthermore, it is not clear whether the r^2s in Fig. 1E are computed in a held-out manner (more details about what goes into computing r^2 are needed). Assessing the model based on peak location alone (Fig. 1E) is not sufficient, as other smooth monotonically-decreasing functions may perform similarly. Simulating from the model greatly improves the reader's understanding (Fig. 2D), but no explanation is given for why the simulations (Fig. 2D) have almost no background spikes and much fewer, non-co-occurring bursts than those of real data (Fig. 2E). A key assumption of the distance model (Fig. 2A) is that each neuron has the same gaussian fall-off (i.e., sigma_excitation and sigma_inhibition), but it is unclear if the data support this assumption. Although an excitatory and inhibitory gain is assumed (Fig. 2A), it is not clear from the data (Fig. 1C) that an inhibitory gain is needed (no negative correlations are observed in Fig. 1C-D). After optimization (Fig. 3), the model is evaluated on predicting burst properties but not evaluated on predicting held-out responses (R^2s or likelihoods), and no other model (e.g., fitting a GLM or a model with only an excitatory gain) is considered. In particular, one may consider a model in which "assemblies" do exist - does such an assembly model lead to better held-out prediction performance? It is unclear why a genetic algorithm (Fig. 1A-C) is necessary versus a grid search; it appears that solutions in Generation 2 (Fig. 3C, leftmost plot, points close to the origin) are as good as solutions in Generation 30 and that the spreads of points across generations do not shrink (as one would expect from better mutations). Given the small number of parameters (7), a grid search is reasonable, computationally tractable, and easier to understand for all readers (Fig. 3A). It is unclear why the excitatory and inhibitory gains of the temporal profiles (Fig. 3I) appear to be gaussian but are formulated as exponential (formula for I_ij^X in Methods). Overall, comparing this model to other possible (similar) models and reporting held-out prediction performance will support the claim that the distance model is a good explanation for trial-to-trial variability.

Data results: Data results were clear and straightforward. However, the explanation was not given for certain results. For example, the relationship between pre-stimulus linear drive and delta R was weak; the examples in Fig. 4C do not appear to be representative of the other sessions. The example sessions in Fig. 4C have R^2=0.17 and 0.19, the two outliers in the R^2 histogram (Fig. 4D). The black trace in Fig. 4D has large variations (e.g., a linear drive of 25 and 30 have a change in delta R of ~0.1 - greater than the overall change of the dashed line at both ends, ~0.08) but the SEMs are very tight. This suggests that either this last fluctuation is real and a major effect of the data (although not present in Fig. 4C) or the SEM is not conservative enough. No null distribution or statistics were computed on the R^2 distribution (Fig. 4C, blue distribution) to confirm the R^2s are statistically significant and not due to random fluctuations. The absence of any background activity in Fig. 6B (e.g., during the rest blocks) is confusing, given that in spontaneous activity many bursts and background activity are present (Fig. 2E). Finally, it appears that the anterior optic tectum contributes to convergent saccades (CS) (Fig. 7E) but no post-saccadic activity is shown to assess how activity changes after the saccade (e.g., plotting activity from 0 to 60). No explanation is given why activity drops ~30 seconds before a convergent saccade (Fig. 7E). No statistical test is performed on the R^2 distribution (Fig. 7H) to confirm the R^2s (with a mean close to R^2=0.01) are meaningful and not due to random fluctuations.

Presentation: A disjointed part of the paper is that for the first part (Figs. 1-3), the focus is on capturing burst activity, but for the second part (Figs. 4-7), the focus is on trial-to-trial variability with no mention of bursts. It is unclear how the reader should relate the two and if bursts serve a purpose for stimulus-evoked activity.

Citations: The manuscript may cite other relevant studies in electrophysiology that have investigated noise correlations, such as:<br /> - Luczak et al., Neuron 2009 (comparing spontaneous and evoked activity).<br /> - Cohen and Kohn, Nat Neuro 2011 (review on noise correlations).<br /> - Smith and Kohn, JNeurosci 2008 (looking at correlations over distance).<br /> - Lin et al., Neuron 2015 (modeling shared variability).<br /> - Goris et al., Nat Neuro 2014 (check out Fig. 4).<br /> - Umakantha et al., Neuron 2021 (links noise correlation and dim reduction; includes other recent references to noise correlations).

if you add spaces, it is easy to read and understand the code.

Importance of vendors providing open source support contribute back to the community

Combine all google fonts into because to avoid repetition of code

perfect code

I didn't know this code. Thank you.

I know nothing about newborns or their neurological functioning. However, I found it interesting how newborns can begin 'breaking the code' of language, at a stage in their lives when they cannot do much for themselves.

• Dominic

For Mac users having issues https://github.com/microsoft/vscode/issues/103855

You can simply use margin: 50px to shorten code whenever all sides are the same

women didn't have rights and because people like Abigail Adams fought for women we have rights now

I think you can comment this. The code below this has the same purpose, giving gradient and adding background image.

Use of comments to indicate which div you are closing, would help you if you need to review your code

The <br /> tag showed warning for me when I tried to validate my code. So I used <br>.. Everything else looks great.

the right thing you have to do

I clicked the link for reFill and thought it looked interesting. Would like to look into this further.

;; makes the next lines of code unreadable.

<samp> is used to highlight code, it may be better to use a simple <span>

• the dynamic structure of the programming langage
• was the dynamic structure
• of the doument

This IS Intentional Programming

• see : https://indyhub.files.fission.name/p/viewer/web/viewer.html?file=https://indyhub.files.fission.name/p/paper%20-%20simonyi%20-%20death%20of%20computer%20languages%20-%201995%20-%20ip.pdf#annotations:8XPqilEIEe2TdP8XAlWuUA

!- claim : NLS was a nascent Intentional Software System - It was not a language work bench - but a system oriented - a collaboration platform designed to bootstrap itself - into existence and continual improvement and in fact - it was kernel based - and at the heart it had a Command Language Interpreter - that allowed thinking and creating software - as a conversational system - driven by a command language - that was used to articulate the intentions - in terms of other intentions, combinations of intentions ultimately - groundrd in primitiver capabilities - that were implemented in machine code for the target system - migration from one machine to the next - system desigtners and devlopers were able to articulate - ideas and means of realizing those ideas in - self structured forms and used the same capabilities - to organize the embodiment of these intentions - in code that when run exhibited the desired behaviour

If you're thinking without writing, you only think you're thinking. —Leslie Lamport.“Thinking Above the Code.” Lecture presented at the Microsoft Research Faculty Summit, Microsoft Research, July 15, 2014. https://www.microsoft.com/en-us/research/video/leslie-lamport-thinking-code/. Timestamp: 03:26

Link to:<br /> https://hypothes.is/a/rvisgFDXEe2s-SuJJGw3cA<br /> https://hypothes.is/a/yEFMHoCkEeyl34fItJe__w

Note that the spoken quote is different from the written quote.

Lamport, Leslie. “Thinking Above the Code.” Lecture presented at the Microsoft Research Faculty Summit, Microsoft Research, July 15, 2014. https://www.microsoft.com/en-us/research/video/leslie-lamport-thinking-code/.

Presentation slides presumably available at https://www.microsoft.com/en-us/research/wp-content/uploads/2016/07/leslie_lamport.pdf

see also: https://www.wired.com/2013/01/code-bugs-programming-why-we-need-specs/

https://www.wired.com/2013/01/code-bugs-programming-why-we-need-specs/

apparently quoted from Guindon: Michigan So Far<br /> see: https://www.goodreads.com/work/quotes/3403680-guindon-michigan-so-far#:~:text=%E2%80%9CWriting%20is%20nature's%20way%20of,how%20sloppy%20your%20thinking%20is.%E2%80%9D

Link to: https://hypothes.is/a/yEFMHoCkEeyl34fItJe__w along with the commentary there. This quote is another example of this fallacy.

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

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

Summary:

In this manuscript, authors establish a glyco-profiling platform for the functional analysis of genes involved in pseudaminic (Pse) and legionaminic (Leg) acid biosynthetic pathways. They used B. subvibroides and C. crescentus specific mutants in pseI and legI genes involved in the Pse and Leg biosynthesis, respectively, and cross-complementation assays with orthologous genes from different bacterial species, analysing motility and flagellin glycosylation. These assays show that Pse and Leg biosynthetic pathways are genetically different and recognize the LegX enzyme as a critical element in the Leg-specific enzymatic biosynthesis. Since that legX orthologous were only identified in the genome of bacteria with Leg biosynthetic pathways, it becomes a good marker to distinguish Leg from Pse biosynthesis pathways and a novel bioinformatic criterion for the assignment and discrimination of these two pathways. Reconstitution of Leg biosynthetic pathway of B. subvibroides in the C. crescentus mutant that lack flagellins, PseI and FlmG, complemented with both flagellin and FlmG of B. subvibroides, identified a new class of FlmG protein glycosyltransferases that modify flagellin with legionaminic acid. Furthermore, the construction of a chimeric FlmG through domain substitutions, allowed to reprogram a Pse-dependent FlmG into a Leg-dependent enzyme and reveal two modular determinants that govern flagellin glycosyltransferase specificity: a glycosyltransferase domain that accepts either Leg or Pse, and a specialized flagellin-binding domain to identify the substrate.

Major comments:

The conclusions obtained are convincing and well-supported. However, I think some points should be specify or clarify.

1.- In the mutants (pseI, legI, flmG,...) the non-glycosylated flagellin are exported and assembled in a flagellum filament shorter than the WT strain. However, motility in plates is absent or very reduced. This might be produced by instability of the flagellum filament when rotating in a semi-solid surface. MET was performed from plates or liquid cultures? Do the author analyses motility in liquid media? If they did, changes in motility were observed?

Response: The Caulobacter ΔpseI mutant accumulates low levels of flagellin in the supernatant. TEM analysis reveals that the flagellar filament is not assembled and only the hook structure is visible (PMID: 33108275). Brevundimonas subvibrioides ΔlegI or ΔflmG cells feature a shorter filament compared to WT by TEM. In all these analyses, TEM was performed on cells grown in broth to exponential growth phase as detailed in the Experimental procedures section. These mutant cells do not swim when analyzed by phase contrast microscopy. While is not known if swimming on semi-solid medium would further destabilize the flagellar structures seen in liquid cultures by TEM, there is more residual motility in B. subvibrioides mutants that make a short filament compared to C. crescentus mutants that lack the flagellar filament. Thus, our analyses point to a positive correlation between the residual motility and residual filament length when comparing the B. subvibrioides and C. crescentus mutants.

2.- In page 5, lines 158-163, the analysis, by HPLC, of derivatized nonulosonic acid from B. subvibroides flagella, shows a major peak at 9.8 minutes retention and a minor peak at 15.3 minutes. Since that Pse-standard have retentions peaks at 9.7 and 13 minutes, and Leg-standard at 12.3 minutes, the authors cannot infer, only with these data, the flagella sugar is a legionaminic acid derivative. In my opinion, should be included that inference comes from the data obtained by HPLC analysis and genetic approaches. Thanks. Corrected. 3.- In page 5, line 173-175. Authors indicate, "While no difference in the abundance of flagellin was observed in extracts from mutant versus WT cells, flagellin was barely detectable in the supernatants of mutant cultures, suggesting flagellar filament formation is defective in these mutants". MET images show that the flagellum filament length is shorter in the mutants than in the WT strain. Therefore, if the same number of mutants and WT cells has been used in the immunodetection assays, there should be more flagellin monomers in the WT samples than in the mutants ones and flagellin bands should be less intense in mutant samples corresponding to the anchored flagellum. Why bands corresponding to flagellin in mutants and WT show similar intensity in the immunodetection assays (Figure 3C and D)? Furthermore, in lane 177-178, authors suggest that LegI and FlmG govern flagellin glycosylation and export (or stability after export). However, if filament stability is affected, the amount of flagellin monomers in the supernatant of mutants should be higher than in the WT. However, immunodetection assays show less abundance of flagelin monomers in the supernatant of mutants. Please, can you clarify this? In relation to this point, I suggest that authors include, in the experimental procedures, how they obtained the supernatants to flagellin immunodetection, as well as why they used anti- FljKCc anti-serum to detect the B. subvibroides flagellin.

We thank the reviewer for raising this point. We have now clarified this question in the updated Experimental procedures section. Our immunoblots harbor the same number of cells harvested in exponential phase (OD=0.4). One mL of cells was harvested from cultures by centrifugation at full speed. The supernatant that was used for the immunodetection corresponds to the supernatant after the centrifugation. The supernatant fraction contains flagella that have been shed during the cell cycle at the swarmer cell to stalked cell (G1-S) transition of C. crescentus and B. subvibrioides.

Thus, it is clear that the majority of flagellins detected by immunoblotting are in fact cell associated and specifically the intracellular flagellins. The evidence for this is that the levels are comparable between WT and ΔflmG mutant cells, even though the latter has shorter or no flagellar filaments. Moreover, while C. crescentus cells are not constantly flagellated during the cell cycle, flagellins are detectable on cell-associated samples by immunoblotting even when cells do not yet or no longer have a flagellar filament. Based on these two points, we conclude that the total flagellin levels associated with cells do not reflect the levels of flagellin assembled into a flagellar filament, but rather the flagellin bulk present in the cytoplasm.

Consistent with this view, we previously reported that C. crescentus ΔpseI cells have the same amount of flagellins in cell lysates compared to the WT strain (PMID: 33108275), even though the mutant cells lack a flagellar filament. Thus, the results obtained here are consistent with previous observations and indicate that B. subvibrioides flagellin glycosylation mutants also still produce comparable amounts of flagellins intracellularly like the WT strain, despite the absence of flagellin glycosylation and inefficient assembly into a flagellar filament.

Concerning the potential role of LegI and FlmG in flagellin stability after export, we were referring to protein stability (half-life), not filament stability. Glycosylation may impact the half-life of extracellular flagellins since glycosylation can protect from proteolytic degradation of proteins, possibly in this case by different proteases that may accumulate in the supernatant. Thus, non-glycosylated flagellins could be more easily degraded by extracellular proteases once they are exported, ultimately resulting in a lower amount in the supernatant.

Addressing the final question about the specificity of the anti-FljKCc antiserum: we used this anti-serum because it detects the B. subvibrioides flagellins owing to the high sequence similarity between B. subvibrioides flagellins and C. crescentus flagellins. We previously showed that the anti-FljKCc anti-serum detects all six flagellins from C. crescentus, as determined by individually expressing each flagellin in a strain deleted for all six flagellin genes (Δfljx6) (PMID: 33108275). FljKCc (against which the antibody was raised) is 65% similar to the most distant C. crescentus flagellin, FljJ. As the similarity of FljKCc to the three B. subvibrioides flagellins ranges from 74% -67% sequence similarity, they should be even better recognized by the anti- FljKCc antibody than C. crescentus FljJ. However, on immunoblots we cannot attribute the signal to any individual B. subvibrioides flagellin as they could all co-migrate on SDS-PAGE and therefore all flagellins might reside in the same immunoblot band. However, we can clearly demonstrate that the immunoblot band corresponds to flagellins: a B. subvibrioides ΔflaF mutant (see below) that we constructed revealed that the flagellin signal is lost, as is the case for a C. crescentus ΔflaF mutant (PMID: 33113346). In the case of C. crescentus, the FlaF secretion chaperone is required for flagellin translation (synthesis) and we suspect that this also the case for B. subvibrioides FlaF. This experiment provides additional evidence that the B. subvibrioides flagellins are recognized by the anti-FljK (C. crescentus) anti-serum.

4.- Authors demonstrate the specificity of the GT-B domain of FlmG, using a chimeric FlmGCc-Bs in a mutant of C. crescentus that lacks FlmG and harbour the Leg biosynthetic pathway of B. subvibroides. However, since that TPR comes from C. crescentus, this chimeric protein, could be transfer the legionaminic acid to the flagellin of B. subvibroides? Furthermore, the complementation of this mutant with the FlmGBs did not support efficient flagellin modification and this might be related to the TPRCc domain. Therefore, in my opinion, the chimeric protein should be introduced in the B. subvibroides∆flmG background. The answer to the first question is “No” or “very inefficiently” as determined from immunoblot analyses of B. subvibrioides ΔflmG cells expressing the chimeric FlmG_Cc-Bs protein that we now show in Fig S2B.

Expression of the different FlmG (FlmG_Cc, FlmG_Bs, FlmG_Cc-Bs) in C. crescentus cells producing Pse or Leg revealed that FlmG_Bs does not support efficient flagellin modification with Pse in C. crescentus, likely because FlmG_Bs interacts poorly with the C. crescentus flagellins. By using the FlmG_Cc-Bs chimera we hoped to overcome this interaction problem with the C. crescentus flagellins (because the FlmG chimera harbors the C. crescentus TPR to bind the C. crescentus flagellins), however glycosyltransfer still does not occur efficiently because the GT domain from FlmG_Bs does not function with Pse. However, FlmG_Cc-Bs can modify the C. crescentus flagellins once C. crescentus is genetically modified to produce CMP-Leg (instead of CMP-Pse). This confirms that the FlmG TPR from C. crescentus is important for flagellin modification through the FlmG/flagellin interaction and that GT_B type glycosyltransferase only transfers Leg. In addition, we have now added as Fig S2B an immunoblot and as Fig S2C a motility test of B. subvibrioides ΔflmG cells expressing the FlmG_Cc-Bs chimeric protein in which we only observed little modification of B. subvibrioides flagellins and a poor motility, respectively. We extended our discussion of these results.

5.- Page 8, line 299-301. Authors point out that C. crescentus that lacks FlmG and harbour the Leg biosynthetic pathway of B. subvibroides and the chimeric FlmGCc-Bs, although it has a glycosylated flagellin, whose mobility in SDS-PAGE is like the WT strain, is non-motile. They suggest that additional factors exist in the flagellation pathway that exhibit specificity towards the glycosyl group that is joined to flagellins. However, would be interesting to see if the flagellum filament has similar length to the WT strain or at least, it has increased in relation to the flagella length of the mutant. If flagella length has not increased, it could suggest that changes in the glycan type might affects the flagellin assembly or the stability of the flagellum filament. Therefore, would be also important to analyse its motility in liquid media.

To investigate why the C. crescentus cells that produce Leg and express the chimeric FlmGCc-Bs glycosyltransferase are non-motile (Figure S5B) despite flagellin modification (by immunoblotting, Figure 7C), we employed two strategies. First, we performed immunoblot analyses on the supernatant fraction from these cells to determine if flagellins accumulate extracellularly. As now showed in Figure S5A, only low amounts of C. crescentus flagellins modified by Leg are present in the SN fraction. Second, we conducted TEM analyses of cells grown to exponential growth phase in broth. As shown in Figure S5C, the C. crescentus cells producing Leg and expressing FlmG_Cc-Bs glycosyltransferase harbor a shorter flagellum compared to those expressing the FlmG_Cc in which C. crescentus flagellins are modified by Pse. Altogether these results explain why these cells are non-motile both on soft agar plate and in liquid.

Minor comments: 1.- Pag 3 line102. Please change ".....two predicted synthases, a PseI and LegI homolog, and C. crescentus only encodes only PseI...." to ".....two predicted synthases, a PseI and LegI homolog, and C. crescentus only encodes a PseI...." 2.- Figure 2 A. Plasmid nomenclature (Plac-neuB) is confusing because C.c. ΔpseI cells express predicted LegI or PseI synthases. Please change to Plac, as in Figure 2B and 4. Figure 2A and 2B do not contain any complementation with Bacillus subtilis (Basu), however two complementation are labelled as Bs in Figure 2A and 2B. Furthermore, no Bs are present in the Figure 2 legend. 3.- Legend of figure 3 should include B. subvibrioides abreviation Bs. Line 774: Please change ".......glycosylation and secretion in B. subvibrioides." to ".......glycosylation and secretion in B. subvibrioides (Bs)." 4.- Figure 3. In order to keep a similar nomenclature in all plasmids, plasmid Plac-legI syn and Plac-flmG should be labelled as Plac-legIBs syn and Plac-flmGBs.

5.- Legend of figure 4 should include B. subvibrioides abreviation Bs. Line 791: Please change "....... complementation of the B.subvibrioides ΔlegI mutant with ...." to "....... complementation of the B.subvibrioides (Bs)ΔlegI mutant with ...." Furthermore, Legend of figure 4 indicate in line 795, that immunoblots reveal the intracellular levels of flagellin, however figure 2 and 3 show immunoblot of cell extracts. Please, correct this sentence. 6.- Legend of figure 5, 6 and 7 should include B. subvibrioides abreviation Bs. Line 808: Please change "Predicted Leg biosynthetic pathway in B. subvibrioides " to"Predicted Leg biosynthetic pathway in B. subvibrioides (Bs)" Line 834: Please change "....affects motility, flagellin glycosylation and secretion in B. subvibrioides."to "....affects motility, flagellin glycosylation and secretion in B. subvibrioides (Bs).Line 852: Please change "...acetyltransferase in flagellar motility of B. subvibrioides cells." to ""...acetyltransferase in flagellar motility of B. subvibrioides (Bs) cells." Furthermore, figure 5 should include C. crescentus abbreviation. Line 815: Please change "....whole cell lysates from C. crescentus mutant cultures......." to "....whole cell lysates from C. crescentus (Cc) mutant cultures......." 7.- In my opinion it would be useful to include a scheme of the gene organization involved in Leg biosynthesis in B. subvibrioides.

8.- Legend of figure S1 should include B. subvibrioides (Bs) and C. crescentus (Cc) abbreviations. Line 888-867: Please change "...C. crescentus ΔpseI cells and B. subvibrioides ΔlegI cells with plasmids expressing..." to "...C. crescentus (Cc) ΔpseI cells and B. subvibrioides (Bs) ΔlegI cells with plasmids expressing..." Furthermore, the name and abbreviations (Mm, So, Ku, Pi, Dv) of the species used should be included in the legend. Why the authors used a plasmid with a Pvan promoter in these assays? Why the authors changed the code color of pseI and legI orthologous genes? It would be more useful and understandable follow the code color used in figure 2 and 4.

Page 6 line 200, Please change ".....complementing synthases exhibit greater overall sequence similarity to LegI than Pse of C. jejuni. 22268,....." to ".....complementing synthases exhibit greater overall sequence similarity to LegI than PseI of C. jejuni. 22268,....." 10.- Page 7 line 231, Please change ".....negative bacteria A. baumannii LAC-4 (GCA_000786735.1)[38] and P. sp. Irchel 3E13..." to ".....negative bacteria A. baumannii LAC-4 (GCA_000786735.1)[38] and Pseudomonas sp. Irchel 3E13..." 11.- Introduce a line break between line 503 and 504. 12.- Page 14 line 543, please change "XbaI" to "XbaI" Thanks for the careful editing. We changed the text as suggested by the reviewer. We also added a scheme showing the genetic organization of the genes involved in Leg production and present as Figure 1B. When this study was initiated, the pMT335 plasmid with a Pvan promoter was used before we switched to using the pSRK plasmid with Plac promoter for better induction. Note that the results with Pvan or Plac are comparable regarding the PseI synthases interchangeability. Color code is now homogenous through the manuscript.

Reviewer #1 (Significance (Required)):

This is an interesting manuscript that contributes to the knowledge of the legionaminic biosynthetic pathway and establish a glyco-profiling platform for the functional analysis of genes involved in pseudaminic (Pse) and legionaminic (Leg) acid biosynthetic pathways. The analysis of Leg patway allowed to identify a gene (legX) that can be used to distinguish Leg from Pse biosynthesis pathways, becoming a bioinformatic tool for the assignment and discrimination of these two pathways. Furthermore, a new class of FlmG protein glycosyltransferases, able to transfer Leg to the flagellin, has been identified and its analysis reveal two modular determinants that govern flagellin glycosyltransferase specificity: a glycosyltransferase domain that accepts either Leg or Pse, and a specialized flagellin-binding domain to identify the substrate.

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

Summary: Viollier and co-workers present a study in which they preform an elegant and rigorous genetic profiling of the the legionaminic and pseudaminic acid biosynthesis and flagellar glycosylation pathways in C. crescentus (native Pse) and B. subvibrioides (native Leg). They use motility as a representative readout for functional flagellar glycosylation with these microbial sialic acids. They discover orthologous Pse synthase genes can replace the function of the native synthase in C. crescentus and orthologous legionaminic acid synthase genes can achieve the same in B. subvibrioides. However, not vice versa indicating a strong preference for each microbial sialic acid stereoisomer in these species. For the Leg biosynthesis pathway, which requires GDP-GlcNAc, the authors also identify LegX as an essential component to synthesize this sugar nucleotide and thus a marker for Leg biosynthesis pathways. Upstream in theses pathways, they also identify a new class of FlmG flagellar protein glycosyltransferases. Importantly, through heterologous reconstitution experiments to uncovered that these glycosyltransferases possess two distinct domains, a transferase domain the determines specificity for either CMP-Leg or CMP-Pse, and a flagellin-binding domain to achieve selectivity for the substrate. Interestingly, by creating chimeric FlmG for these two domains between C. crescentus and B. subvibrioides they show that these two modular parts can be interchanged to adapt flagellin glycosyltransferase specificity in these species. Major comments: The key conclusions of the manuscript by Viollier and co-workers are convincing and well supported by their experiments and used methods, with respect to the insulation of the Leg and Pse biosynthetic pathways, they key role of LegX in launching the Leg pathway and the successful reconstitution of Leg glycosylation in a previously Pse-producing C. crescentus strain. Finally, they convincingly show that a chimeric version of the involved glycosyltransferases is functional, which besides intriguing future glycoengineering possibilities also emphasizes the two discrete domains in these transferases that dictate their sugar nucleotide and acceptor specificity. There is one additional experiment I would suggest with relation to the detection and confirmation of Pse and Leg on flagella of respectively, C. crescentus and B. subvibrioides. In the case of C. crescentus the detected DMB derivatized monosaccharide co-elutes with a validated standard of tri-acetylated Pse, which is convincing evidence of its identity. However, for B. subvibrioides. Their DMB derivatized monosaccharides from its flagella, results in a peak the does not co-elute with the only Leg standard (Leg5Ac7Ac) they have, it does elute at the same time as their Pse standard. Although it cannot of course be Pse as B. subvibrioides. Does not possess a Pse biosynthesis pathway, it also does not provide enough evidence to conclude that it is a Leg derivative. An MS(-MS) measurement of the eluted signal would not be a big investment in time and resources and would provide additional evidence to at least assign this peak to microbial sialic acid related to the present Leg biosynthesis pathway. It the identified mass would lead to identification of the derivative, it would also add to the proper characterization of the flagella glycosylation in the bacterium.

We have now added the glycopeptide analyses as requested. They are described in the last experimental section and confirm our results.

The data and the methods presented in this study are presented with sufficient detail so that they can be reproduced? However, I would suggest as is common nowadays in most journals that the authors include images of the raw unprocessed blot in de supporting info.

*The motility pictures are representative of three independent experiments and the immunoblots are representative of at least two independent experiments. This has now been mentioned in the Experimental procedures. The raw unprocessed blots have now been added as supporting info. *

Minor comments: There are a few textual errors that the authors should fix: -page 2, line 70: change "used" to "use" -page 11, line 407: add the word "are" after Pse On page 2, line 36, the authors state that "most eubacteria and the archaea typically decorate their cell surface structures with (5-, 7-)diacetamido derivatives, either pseudaminic acid (Pse) and/or its stereoisomer legionaminic acid (Leg,". This should be nuanced as to my knowledge it is not most eubacteria, but more a subset as identified by Varki in his seminal PNAS paper. The authors clearly present their data and conclusions in the figures of this manuscript. However, I would recommend the take a critical look at the drawing of their monosaccharide chair conformations and the positioning of the axial and equatorial groups on these chairs in Figure 1 and 5, as these are in most cases drawn a bit crooked, which can easily be corrected. We corrected the text as the reviewer suggested. We changed the sentence in the introduction to be more nuanced. The drawing of the monosaccharide has been improved.

Reviewer #2 (Significance (Required)):

The family of carbohydrates called sialic acids was long thought to exclusively occur in glycoproteins and glycolipids of vertebrates, but has since also been found in specific microbes. Especially symbiotic and pathogenic microbes associated with the humans express a wide array of unique microbial sialic acids for which their functional roles are not well understood and the associated glycosylhydrolase and glycosyltransferase have in most cases not been identified yet. The authors present an impressive insight into flagellar glycosylation with Pseudaminic and Legionaminic acid in two bacterial species, using genomic analysis, rewiring, immunoblots and motility assays as their main tools. They provide compelling evidence on the insulation of the Pse of Leg pathway in these species, the flexibility in exchanging between biosynthetic enzymes from the same pathway between various species. Crucially, most glycosyltransferases that add the Pse or Leg glycoform onto various acceptor sites in bacteria, have up to this point remained elusive in most cases. It is therefore very valuable information that the authors here provide on the involved glycosyltransferases. Especially, on the two domains that govern their sugar nucleotide and acceptor specificity, and that these can be reengineered as chimeric glycosyltransferases. To me as a chemical glycobiologist this provides compelling possibilities for glycoengineering possibilities in future studies in the field to elucidate the functional roles of Pse and Leg glycosylation.

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

Summary of the findings and key conclusions (including methodology and model system(s) where appropriate): Kint et al describe a neat study of bacterial flagellin glycosylation by a recently identified class of protein glycosyltransferases called FlmG. The experiments are well designed, the data presented is convincing and the conclusions drawn are mostly in line with the experimental evidence presented. These are the key findings. Kint et al show that genetic tools and motility can be used as a readout to probe the sugar biosynthesis pathway in bacteria. Using the recently characterized system of Caulobacter crescentus, they have performed a survey of different PseI/LegI/NeuB genes from various bacteria, checking whether they could rescue the motility defect in C. crescentus ΔpseI cells. They found that those genes that did confer motility also had higher sequence similarity to C. jejuni PseI than to C. jejuni LegI or C. jejuni NeuB. They also found that these genes also restored flagellin glycosylation as checked by mobility shift on gel electrophoresis with immunoblotting to anti-FljK antibody. This survey brought up an interesting finding that the PseI/LegI/NeuB orthologs of the closely related Brevundimonas species were unable to confer motility to C. crescentus ΔpseI cells, and were more similar to C. jejuni LegI than to C. jejuni PseI or C. jejuni NeuB. They also performed similar glycoprofiling experiments using B. subvibrioides ΔlegIBs cells and various PseI/LegI/NeuB orthologs from different bacteria, which indicated the restoration of motility by putative LegI synthases. Kint et al demonstrate flagellin glycosylation in B. subvibrioides by performing in-frame deletions of FlmG, and LegI genes in B. subvibrioides and checking for motility, presence of flagella, and flagellin glycosylation by motility shift on gel electrophoresis. Further, they confirm the critical nature of GDP-GlcNAc for Leg biosynthesis by assessing flagellin glycosylation and motility in B. subvibrioides with an in-frame deletion in PtmE/LegX and by performing heterologous complementation with an M. humiferra PtmE ortholog. They also reconstitute the legionaminic acid biosynthesis pathway in C. crescentus cells that lack flagellins, PseI and FlmG, and show that the heterologously expressed B. subvibrioides flagellin is glycosylated by heterologously expressed B. subvibrioides FlmG. Finally, they also show that whereas the CcFlmG cannot substitute for BsFlmG and vice versa, a chimeric FlmG bearing the TPR domain from C. crescentus FlmG (that recognizes C. crescentus FljK) and the GT domain from B. subvibrioides FlmG (that transfers CMP-Leg) modifies CcFljK in C. crescentus cells that lack CcFlmG but express both Pse (endogenously) and Leg (from the reconstituted pathway). This demonstrates the modularity of the FlmG glycosyltransferases. Kint et al provide the chemical nature of C. crescentus flagellin glycosylation. Kint et al have analyzed the glycans released from the flagellin by acid hydrolysis and clearly shown the nature of the glycan in C. crescentus flagellin to be Pse4Ac5Ac7Ac by use of Pse standards. The glycan from B. subvibrioides was distinct from the Leg standard used, and could be a Leg derivative distinct from Leg5Ac7Ac.

Major comments: 1. Table 1 and Text in Results, lines 116-119, "In support of the notion that derivatization occurs after the PEP-dependent condensation reaction to form Pse or Leg, our glyco profiling analysis revealed that putative PseI proteins (identified by sequence comparisons to C. jejuni 11168, Table S1) conferred motility to C. crescentus ΔpseI cells, whereas putative LegI synthases did not." Not clear how putative PseI and LegI synthases were identified. Table 1 only lists overall percent sequence identity and similarity to Cj PseI, LegI and NeuB, and percent identities and similarities of the various nonulosonic synthases to these proteins are in the similar range, as expected. In the absence of sequence alignments indicating the presence of conserved residues, particularly related to the substrate binding region, that are distinct in these paralogs, calling out the type of synthase based on the highest percent identity (to Cj PseI, LegI or NeuB) is speculative. Also, Shewanella oneidensis does not follow the pattern of highest similarity to NeuB3. Second, in the absence of data showing that the Leg and Pse found in these different organisms actually are different derivatives, this does not support that "derivatization occurs after the PEP-dependent condensation reaction to form Pse or Leg". Putative PseI and LegI were proposed based on BlastP analyses in which the protein sequences of interest were aligned to the three experimentally validated synthases from C. jejuni 11168: PseI, LegI, NeuB as well as PseI from C. crescentus, as indicated in Table S1. While, the assignment of the donor sugar is based only on the sequence identity and similarity to LegI or PseI, this assignment corresponds well according to the restoration of the motility of the C. crescentus ΔpseI mutant upon expression of PseI ortholog and B. subvibrioides ΔlegI mutant with heterologous LegI expression.

It is true that for Shewanella oneidensis the assignment as PseI or LegI is ambiguous, exhibiting nearly identical similarity, but it is quite distinct from NeuB. This actually makes the S. oneidensis synthase a very interesting case to explore the enzymology of its Pse/LegI ortholog, knowing that it has been previously shown that this bacterium glycosylates its flagellins with Pse derivatives (PMID: 24039942). The results from our genetic complementation analysis are however very clear (PseI ortholog) and very consistent with the functional analysis in S. oneidensis.

Concerning the different derivatives of Pse or Leg: McDonald and Boyd (PMID 32950378) recently published a review giving some examples of Bacteria/Archaea experimentally shown to contain Pse/Leg-derivatives: C. jejuni 11168 modifies its flagellin with 5,7-N-acetyl Pse, Sinorhizobium fredii NGR234 (not used in this study but in our previous work PMID 33113346 and showed to restore the motility of C. crescentus ΔpseI cells) modifies its capsule with 5-acetamido-7-3-hydroxybutyramido-Pse), Treponema denticola modifies its flagellin with 7-(2-metoxy-4,5,6-trihydroxy-hexanoyl-Pse, A. baumannii LAC-4 produces 5,7-N-acetyl-8-epi-Leg to decorate the capsule, Halorubrum sp. PV6 modifies the LPS with N-formylated Leg and L. pneumophila produces 5-acetamidino-Leg.

The reviewer is right in that we do not know the exact version of Pse or Leg produced in C. crescentus and B. subvibrioides, HOWEVER, the fact that complementation works with the majority of the orthologs of PseI and LegI including many from bacteria that are known to produce modified Pse derivatives for example in Shewanella oneidensis and Treponema denticola, the most likely explanation is that derivatization occurs after the PseI or LegI step, but we concede that the results are also compatible with a promiscuous enzyme that can accept different Pse derivatives or different Leg derivatives.

1. Related to (1), Text in Results, lines 130-131, "We conclude from our survey that (heterologous) PseI synthase activity generally confers motility to C. crescentus ΔpseI cells, whereas LegI-type (or NeuB-type) synthases are unable to do so." There is no a priori evidence provided indicating that these were PseI or LegI type synthases. So the conclusion really is that assuming only PseI type synthases would be able to rescue the motility defect in C. crescentus ΔpseI cells, this glyco-profiling motility assay now provides the first biochemical evidence telling us which synthases are Pse-type, and which are Neu/Leg-type. And in my view, this is the conclusion of greater significance in the field - to be able to now identify which is a PseI and which a LegI based on these complementation assays. However, if the authors still wish to retain their original conclusion, they could cite or provide evidence (either biochemical evidence in this work or reported literature regarding the sugar synthesized or bioninformatics analysis regarding the presence of distinct genes such as the Ptm genes for legionaminic acid biosynthesis pathway or genes that differ in their enzyme activities and overall fold such as PseB/LegB or PseG/LegG in the gene neighborhood) indicating or suggesting the PseI/LegI/NeuB nature of the different synthases. Also, methods for the bioinformatics analysis (eg. BLASTp settings used, dates of searches, whether regular BLAST or PSI-BLAST was used, etc.) are missing in the manuscript, and need to be included. We agree that for many PseI or LegI tested, there is no provided biochemical evidence. HOWEVER, this is not the case for some of them including the PseI, LegI and NeuB from Campylobacter jejuni (PMID 19282391), some A. baumannii strains (α-epi-legionaminic acid for A. baumannii LAC-4 PMID 24690675), Shewanella oneidensis (Pseudaminic acid with methylation PMID 23543712), Legionella pneumophila (Legionaminic acid PMID 18275154) or Halorubrum sp. PV6 (N-formylated legionaminic acid PMID 30245679). Thus, we maintain the two conclusions: the PseI and LegI synthases are generally interchangeable and the complementation assays can enable to identify and assign PseI and LegI function. BLAST2P was used to compare the protein sequences of the tested NeuB-like synthases with NeuB1, LegI (NeuB2) and PseI (NeuB3) from Campylobacter jejuni but also with PseI from C. crescentus. BLOSUM62 matrix was used as well as a word of size 3 for the comparison. We have now added this procedure in the legend of the Table S1.
2. It is interesting that there is still a signification amount of flagellin secretion/assembly in the B. subvibrioides LegI and FlmG mutants. It will be good to see a discussion about whether this is likely from due to low level of function despite the in-frame deletion of genes; how many flagellin subunits are likely to have managed secretion and assembly in these short flagella; whether there is any redundancy of LegI / FlmG (perhaps with lower levels of expression); considering Parker and Shaw's findings of glycosylation being required for flagellin binding to the chaperone and subsequence secretion in A. caviae whether there is a FlaJ homolog in B. subvibrioides. Also, can the authors rule out the possibility that absence of glycosylation does not affect flagellin assembly but makes the flagellum prone to shear/breaks in B. subvibriodes, resulting in smaller flagella? How many flagellins are there in B. subvibrioides? Is it possible that one is glycosylated but another/others are not, and that is the reason for the small flagellum in these mutants? The number of flagellin subunits that are assembled into a full-length flagellar filament is unknown in C. crescentus and in B. subvibrioides. There are 3 different flagellin genes that are now presented schematically in Figure 1C. No redundancy has been found for LegI or FlmG. It is possible that the B. subvibrioides is better in exporting non-glycosylated flagellin or that the capping proteins can function better with sugar modification or that the filament of B. subvibrioides mutants is less fragile when it is non-glycosylated or that its flagellins “stick” better. It is also possible that short filaments are not actually containing flagellins mounted on the hook but another protein that polymerizes aberrantly in the absence of Leg or FlmG. This remains to be investigated and compared to the situation of Pse and FlmG mutants of C. crescentus.

B. subvibrioides possesses an ortholog of the C. crescentus flagellin secretion chaperon FlaF (PMID 33113346). As observed in C. crescentus, FlaF likely has a role in flagellin translation as its inactivation totally prevents flagellins production (see answer to reviewer #1). For C. crescentus, bacterial two hybrid experiments revealed that FlaF can interact with non-glycosylated flagellins in E. coli. Thus, it is strongly possible that FlaF/flagellins interaction is not dependent on the flagellins glycosylation state. In addition, the short flagellum filament observed in B. subvibrioides ΔlegI or ΔflmG mutants argues that at least some flagellins are secreted while not glycosylated.TEM pictures have been performed in liquid medium from exponential growth phase. In this condition, no fragment of flagella was observed in the culture medium by TEM but only small flagella with a hook structure attached. Also, flagella breaks might result in more random length of flagellum.

Three flagellins are in B. subvibrioides (Bresu_2403 is 59% identical with FljLCc, Bresu_2638 is 57% identical with FljKCc and Bresu_2636 is 62% identical with FljJCc). We now show this genetic organization of the flagellins in Fig. 1C. The three flagellins are all detected by the anti-FljKCc anti serum (see answer and figure to reviewer #1). We cannot attribute the immunoblot signal to any individual B. subvibrioides flagellin as they could all co-migrate on SDS-PAGE. However, the signal often looks like a doublet (as shown in Figure 4B for example) suggesting that at least two flagellins are detected and this doublet is always found to migrate faster in absence of glycosylation that could indicate that all B. subvibrioides flagellins (or at least 2) are modified.

Text in Results, lines 170-171, "We then probed the resulting ΔlegIBs and ΔflmGBs single mutants for motility defects in soft agar and analyzed flagellin glycosylation by immunoblotting using antibodies to FljKCc". Was the antibody to FljKCc determined to also specifically bind to FljKBs? Also, how many flagellins are there in B. subvibrioides? Are all detected with this antibody? Antibodies raised to FljKCc were raised against His6-FljK produced in E. coli (previously published in Ardissone et al, 2020). This serum recognizes the 6 flagellins from C. crescentus (PMID: 33108275). It recognized the three flagellins from B.s. (see answer to reviewer #1).

It is interesting that C. cresentus cells expressing Pse (endogenously) and Leg (reconstituted pathway), and BsFlmG and BsFljK (corresponding to Figure 5C) are not motile. Was the motility assay done for the experiment of figure 5B as well? Are the C. crescentus cells lacking Pse and FlmG but with heterologous expression of Leg and BsFljK and BsFlmG also non-motile? Also, it will be good to see the TEM images for these cells.

C. crescentus cells that produce Pse (endogenously) or Leg (reconstituted pathway) and BsFlmG and BsFljK (formerly Figure 5C and now as Figure 7C) are indeed not motile as shown by the motility tests presented in Figure S5B. Motility assays with cells used in the former Figure 5B (now Figure 7B) have also been done and are now presented Figure S4B. These cells are non-motile because BsFljK is not efficiently secreted (or unstable after secretion) as shown on the immunoblot of the supernatant fraction in Figure S4A lower panel. As a result, flagellar filament is not properly assembled as only a short flagellum was observed by TEM in such cells compared to the WT C. crescentus (Figure S4C and S4D).

Immunoblotting of the supernatants should be shown (in addition to the cell extracts) for Figures 5B and 5C so that the reader can appreciate whether glycosylation has taken place but secretion/assembly has not. Further, HPLC of the acid extracts from flagellin could be done to unambiguously show whether the CcFlmG has transferred Pse and the BsFlmG and Cc-BsFlmG have transferred Leg on to the CcFljK in Figure 5c, and the identity of the sugar, if any, transferred by CcFlmG in the absence of Pse, and BsLeg genes or BsLegX gene in figure 5B.

*__ Immunoblots of the supernatants for Figure 5B (now Figure 7B) have been done and been added (Figure S4A lower panel). BsFljK is barely detected in the supernatant whatever its glycosylation state (with or without Leg). Note that in the supporting info where the raw unprocessed blot used for this panel is shown, a positive control of blotting (C. crescentus Δfljx6 mutant expressing CcFljK from pMT463) has been used. Immunoblots of the supernatant from Figure 5C (now 7C) have been done and been added in figure S5A. The CcFljK modified with Leg is poorly secreted (or unstable after secretion). As a result, these cells only harbor a short flagellum compared to those that are able to modify CcFljK with Pse (Figure S5C).

HPLC of the acid extracts from flagellins have been performed on purified flagella obtained by ultracentrifugation. As C. crescentus cells expressing BsFlmG and Cc-BsFlmG harbor no or short flagellar filament, the purification by ultracentrifugation is limited. Thus, to further confirm that CcFlmG has transferred Pse and Cc-BsFlmG (and BsFlmG) has transferred Leg on CcFljK (former Figure 5C and now Figure 7C), we performed immunoblots on the cell extracts of C. crescentus ΔflmG ΔpseI cells that cannot produce Pse but able to produce Leg (reconstituted pathway). These experiments, now presented in Figure 7C (lower panel) confirmed that no modification of CcFljK was observed in C. crescentus cells expressing CcFlmG whereas CcFljK is modified in C. crescentus expressing Cc-BsFlmG, confirming that Cc-BsFlmG has transferred Leg (the only NulO produced in this condition).__*

Text in discussion, lines 334-338, "By extension, having recognized the LegX/PtmE enzyme as a critical element in the Leg-specific enzymatic biosynthesis step (Figure 6) likewise offers another functional, but also a novel bioinformatic, criterion for the correct assignment and discrimination of predicted stereoisomer biosynthesis routes residing in ever-expanding genome databases" It will be nice to see a discussion on the prevalence of PtmE versus GlmU (or equivalent gene), PtmF, PtmA, PgmL in the Leg synthesizing organisms. Is the PtmE but not the other genes found in all cases, which makes it better as a molecular determinant for bioinformatics predictions of the type of pathway? Also, on whether PtmE has any homology to genes in other pathways (not associated with flagellin glycosylation) and how reliable a marker it is to differentiate Leg biosynthesis from Neu5Ac biosynthesis pathways.

GlmU is a potential bifunctional UDP-N-acetylglucosamine diphosphorylase/glucosamine-1-phosphate N-acetyltransferase that can be part of both Pse and Leg pathway (PMID 19282391). Accordingly, a GlmU ortholog is found in C. crescentus and B. subvibrioides that we showed are producing Pse and Leg, respectively. Thus, GlmU cannot be attributed to a Leg pathway signature. On the other hand, PtmE is barely found in the organisms from which PseI orthologs restore the motility of C. crescentus ΔpseI cells.

PtmF, PtmA, PgmL and GlmS are proposed to act upstream of the production of GlcN-1-P that is a precursor of both UDP-GlcNAc and GDP-GlcNAc, the precursors of Pse and Leg respectively. In addition, orthologs of these genes are not prevalent in the Leg synthetizing organisms present in Table S2 using BlastP analyses with C. jejuni proteins as templates.PtmE ortholog is found in most of the Leg synthetizing organisms as shown in Table S2 and often genetically linked with other genes coding for proteins involved in Leg production (shown with the asterisk * in table S2). Of note, PtmE is found not only in organisms that modify flagellin(s) with Leg but also in organisms that add Leg on capsule such as A. baumannii LAC-4.

It is not clear from the methods or the figure legends how many times the immunoblotting, motility experiments were done; how many experiments/trials are the images representative of? The motility pictures are representative of three independent experiments. The immunoblots are representative of at least two independent experiments. This information is now added in the Experimental procedures section.

Minor comments:

1. The gene for GlcN-1-P guanylyltransferase in the Leg-specific enzymatic biosynthesis step is already known as PtmE from the work of Schoenhofen's group. For the sake of consistency, it would be better to retain the nomenclature as PtmE throughout the manuscript instead of introducing the name LegX, which makes it sound like it is a previously unknown gene.

2. Text in abstract, lines 15-17: "Sialic acids commonly serve as glycosyl donors, particularly pseudaminic (Pse) or legionaminic acid (Leg) that prominently decorate eubacterial and archaeal surface layers or appendages" The glycosyl donor is the nucleotide sugar and not the nonulosonic acid or sialic acid... rephrasing required for accuracy. Done

3. Text in abstract, lines 18: "a new class of FlmG protein glycosyltransferases that modify flagellin" The authors are presumably referring to FlmG as the new class of protein glycosyltransferases... rephrasing required for accuracy Corrected
4. Text in introduction, lines 41-42 "Pse and Leg derivatives synthesized in vitro can be added exogenously in metabolic labeling experiments" It should be "derivatives of Pse and Leg precursors" and not "Pse and Leg derivatives" corrected

5. Text in introduction, line 46 "Pse- or Leg-decorated flagella may also be immunogenic." This sentence is not referenced and it is not clear why it is written here.

6. Text in introduction, lines 63-66 "The synthesis of CMP-Pse or CMP-Leg proceeds enzymatically by series of steps [20-22], ultimately ending with the condensation of an activated 6-carbon monosaccharide (typically N-acetyl glucosamine, GlcNAc) with 3-carbon pyruvate (such as phosphoenolpyruvate, PEP) by Pse or Leg synthase paralogs, PseI or LegI, respectively" The synthesis begins with activated GlcNAc. The substrate for condensation is not activated GlcNAc. It is 2,4-diacetamido-2,4,6-trideoxy-D-mannopyranose in case of LegI and 2,4-diacetamido-2,4,6-trideoxy-b-L-altropyranose in case of PseI. Indeed, we modified the sentence.

7. Text in introduction, line 70 "for used as glycosyl donors" Typographical error, "for use as glycosyl donors" Corrected
8. Text in Results, line 102, "C. crescentus only encodes only PseI" Do the authors mean "only one PseI"? Corrected
9. Text in Results, lines 108 and 109, "Such modifications could occur before the PseI synthase acts or afterwards. In the latter case, most (if not all) synthases would be predicted to produce the same Pse molecule," Do the authors know of any reports of modifications occurring before the PseI synthase? Please cite references, if known. Why "most (if not all)"? If the former case is true, the PseI synthase might not be able to accept the substrate. Correct. Because we cannot test all enzymes we must keep the statement non-committing.

“Most (if not all)” refers to the latter case i.e. the modification occurs after PseI synthase. In this context, PseI should do the same reaction, however, there might be some exceptions.

There is, to our knowledge, no reports showing that modifications occur before the PseI synthase. The glyco-profiling experiments all suggest that modification occurs after Pse production based on our motility readout. It is possible that PseI enzymes that condense a modified precursor would not be functional in our motility assay.

Text in Results, lines 141-143, "our bioinformatic searches using C. jejuni 11168 as reference genome identified all six putative enzymes in the B. subvibrioides ATCC15264 genome (CP002102.1) predicted to execute the synthesis of Leg from GDP-GlcNAc" Not clear how this was done. Do the authors mean that they used the genes from C. jejuni 11168 as the query genes to identify homologs in B. subvibrioides ATCC15264 genome (CP002102.1)? Or did they use putative genes from B. subvibrioides ATCC15264 genome (CP002102.1) and pull out homologs from C. jejuni 11168 by using C. jejuni 11168 as the reference genome? We now have modified the sentence to make it clearer.

At first reading, the flow of the manuscript is difficult to follow due to the figures not appearing in full in order of their occurrence. For instance, Figures 5B and 5C are discussed only in the end of the manuscript after the results of Figures 6 and 7. Other instances also exist. The authors may consider re-ordering the figure parts if possible so that all parts of each figure appear in order of occurrence in the manuscript text. Thanks for raising this issue. We have now tried to address this concern by re-organizing the order of occurrence of the figures. Notably we have now exchanged Figure 5 (on Leg pathway reconstitution and FlmG rewiring) with Figure 7 (on LegB and LegH). We modified the text accordingly. We hope that it makes the manuscript and corresponding figures easier to follow.

Reviewer #3 (Significance (Required)):

The nonulosonic acids, Pseudaminic acid and Legionaminic acid, are abundant in bacterial systems in the capsular and lipopolysaccharides as well as in glycoprotein glycans. The Ser/Thr-O-nonulosonic acid glycosylation of flagellins has been studied with respect to the system of Maf glycosyltransferases in Campylobacter jejuni, C. coli, Helicobacter pylori, Aeromonas caviae, Magnetospirillum magneticum, Clostridium botulinum and Geobacillus kaustophilus, and recently with respect to the system of FlmG glycosyltransferases by Viollier's group in Caulobacter crescentus. However, the determinants that govern the glycosyltransferase function are not still well known. Kint et al have performed excellent work using bacterial genetics tools to (1) highlight the "functional insulation" of the Leg and Pse biosynthesis pathways, (2) demonstrate the modularity of the FlmG glycosyltransferase proteins with respect to the flagellin binding and glycosyltransferase domains. This work makes a significant advance in the field with respect to (1) understanding flagellin glycosylation by FlmG, (2) making designer protein Ser/Thr-O-glycosyltransferases, and (3) bioinformatics analysis of genomes with respect to the Pse/Leg/Neu nonulosonic acid biosynthetic potential encoded. The findings will be of great interest to scientific audiences working in the areas of glycobiology and bacteriology. My area of expertise: Maf flagellin glycosyltransferases

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

Summary:

In this manuscript, authors establish a glyco-profiling platform for the functional analysis of genes involved in pseudaminic (Pse) and legionaminic (Leg) acid biosynthetic pathways. They used B. subvibroides and C. crescentus specific mutants in pseI and legI genes involved in the Pse and Leg biosynthesis, respectively, and cross-complementation assays with orthologous genes from different bacterial species, analysing motility and flagellin glycosylation. These assays show that Pse and Leg biosynthetic pathways are genetically different and recognize the LegX enzyme as a critical element in the Leg-specific enzymatic biosynthesis. Since that legX orthologous were only identified in the genome of bacteria with Leg biosynthetic pathways, it becomes a good marker to distinguish Leg from Pse biosynthesis pathways and a novel bioinformatic criterion for the assignment and discrimination of these two pathways. Reconstitution of Leg biosynthetic pathway of B. subvibroides in the C. crescentus mutant that lack flagellins, PseI and FlmG, complemented with both flagellin and FlmG of B. subvibroides, identified a new class of FlmG protein glycosyltransferases that modify flagellin with legionaminic acid. Furthermore, the construction of a chimeric FlmG through domain substitutions, allowed to reprogram a Pse-dependent FlmG into a Leg-dependent enzyme and reveal two modular determinants that govern flagellin glycosyltransferase specificity: a glycosyltransferase domain that accepts either Leg or Pse, and a specialized flagellin-binding domain to identify the substrate.

Major comments:

The conclusions obtained are convincing and well-supported. However, I think some points should be specify or clarify.

1. • In the mutants (pseI, legI, flmG,...) the non-glycosylated flagellin are exported and assembled in a flagellum filament shorter than the WT strain. However, motility in plates is absent or very reduced. This might be produced by instability of the flagellum filament when rotating in a semi-solid surface. MET was performed from plates or liquid cultures? Do the author analyses motility in liquid media? If they did, changes in motility were observed?
2. • In page 5, lines 158-163, the analysis, by HPLC, of derivatized nonulosonic acid from B. subvibroides flagella, shows a major peak at 9.8 minutes retention and a minor peak at 15.3 minutes. Since that Pse-standard have retentions peaks at 9.7 and 13 minutes, and Leg-standard at 12.3 minutes, the authors cannot infer, only with these data, the flagella sugar is a legionaminic acid derivative. In my opinion, should be included that inference comes from the data obtained by HPLC analysis and genetic approaches.
3. • In page 5, line 173-175. Authors indicate, "While no difference in the abundance of flagellin was observed in extracts from mutant versus WT cells, flagellin was barely detectable in the supernatants of mutant cultures, suggesting flagellar filament formation is defective in these mutants". MET images show that the flagellum filament length is shorter in the mutants than in the WT strain. Therefore, if the same number of mutants and WT cells has been used in the immunodetection assays, there should be more flagellin monomers in the WT samples than in the mutants ones and flagellin bands should be less intense in mutant samples corresponding to the anchored flagellum. Why bands corresponding to flagellin in mutants and WT show similar intensity in the immunodetection assays (Figure 3C and D)? Furthermore, in lane 177-178, authors suggest that LegI and FlmG govern flagellin glycosylation and export (or stability after export). However, if filament stability is affected, the amount of flagellin monomers in the supernatant of mutants should be higher than in the WT. However, immunodetection assays show less abundance of flagelin monomers in the supernatant of mutants. Please, can you clarify this? In relation to this point, I suggest that authors include, in the experimental procedures, how they obtained the supernatants to flagellin immunodetection, as well as why they used anti- FljKCc anti-serum to detect the B. subvibroides flagellin.
4. • Authors demonstrate the specificity of the GT-B domain of FlmG, using a chimeric FlmGCc-Bs in a mutant of C. crescentus that lacks FlmG and harbour the Leg biosynthetic pathway of B. subvibroides. However, since that TPR comes from C. crescentus, this chimeric protein, could be transfer the legionaminic acid to the flagellin of B. subvibroides? Furthermore, the complementation of this mutant with the FlmGBs did not support efficient flagellin modification and this might be related to the TPRCc domain. Therefore, in my opinion, the chimeric protein should be introduced in the B. subvibroides∆flmG background.
5. • Page 8, line 299-301. Authors point out that C. crescentus that lacks FlmG and harbour the Leg biosynthetic pathway of B. subvibroides and the chimeric FlmGCc-Bs, although it has a glycosylated flagellin, whose mobility in SDS-PAGE is like the WT strain, is non-motile. They suggest that additional factors exist in the flagellation pathway that exhibit specificity towards the glycosyl group that is joined to flagellins. However, would be interesting to see if the flagellum filament has similar length to the WT strain or at least, it has increased in relation to the flagella length of the mutant. If flagella length has not increased, it could suggest that changes in the glycan type might affects the flagellin assembly or the stability of the flagellum filament. Therefore, would be also important to analyse its motility in liquid media.

Minor comments:

1. • Pag 3 line102. Please change ".....two predicted synthases, a PseI and LegI homolog, and C. crescentus only encodes only PseI...." to ".....two predicted synthases, a PseI and LegI homolog, and C. crescentus only encodes a PseI...."
2. • Figure 2 A. Plasmid nomenclature (Plac-neuB) is confusing because C.c. ΔpseI cells express predicted LegI or PseI synthases. Please change to Plac, as in Figure 2B and 4. Figure 2A and 2B do not contain any complementation with Bacillus subtilis (Basu), however two complementation are labelled as Bs in Figure 2A and 2B. Furthermore, no Bs are present in the Figure 2 legend.
3. • Legend of figure 3 should include B. subvibrioides abreviation Bs. Line 774: Please change ".......glycosylation and secretion in B. subvibrioides." to ".......glycosylation and secretion in B. subvibrioides (Bs)."
4. • Figure 3. In order to keep a similar nomenclature in all plasmids, plasmid Plac-legI syn and Plac-flmG should be labelled as Plac-legIBs syn and Plac-flmGBs.
5. • Legend of figure 4 should include B. subvibrioides abreviation Bs. Line 791: Please change "....... complementation of the B.subvibrioides ΔlegI mutant with ...." to "....... complementation of the B.subvibrioides (Bs)ΔlegI mutant with ...." Furthermore, Legend of figure 4 indicate in line 795, that immunoblots reveal the intracellular levels of flagellin, however figure 2 and 3 show immunoblot of cell extracts. Please, correct this sentence.
6. • Legend of figure 5, 6 and 7 should include B. subvibrioides abreviation Bs. Line 808: Please change "Predicted Leg biosynthetic pathway in B. subvibrioides " to"Predicted Leg biosynthetic pathway in B. subvibrioides (Bs)" Line 834: Please change "....affects motility, flagellin glycosylation and secretion in B. subvibrioides."to "....affects motility, flagellin glycosylation and secretion in B. subvibrioides (Bs).Line 852: Please change "...acetyltransferase in flagellar motility of B. subvibrioides cells." to ""...acetyltransferase in flagellar motility of B. subvibrioides (Bs) cells." Furthermore, figure 5 should include C. crescentus abbreviation. Line 815: Please change "....whole cell lysates from C. crescentus mutant cultures......." to "....whole cell lysates from C. crescentus (Cc) mutant cultures......."
7. • In my opinion it would be useful to include a scheme of the gene organization involved in Leg biosynthesis in B. subvibrioides.
8. • Legend of figure S1 should include B. subvibrioides (Bs) and C. crescentus (Cc) abbreviations. Line 888-867: Please change "...C. crescentus ΔpseI cells and B. subvibrioides ΔlegI cells with plasmids expressing..." to "...C. crescentus (Cc) ΔpseI cells and B. subvibrioides (Bs) ΔlegI cells with plasmids expressing..." Furthermore, the name and abbreviations (Mm, So, Ku, Pi, Dv) of the species used should be included in the legend. Why the authors used a plasmid with a Pvan promoter in these assays? Why the authors changed the code color of pseI and legI orthologous genes? It would be more useful and understandable follow the code color used in figure 2 and 4.
9. Page 6 line 200, Please change ".....complementing synthases exhibit greater overall sequence similarity to LegI than Pse of C. jejuni. 22268,....." to ".....complementing synthases exhibit greater overall sequence similarity to LegI than PseI of C. jejuni. 22268,....."
10. • Page 7 line 231, Please change ".....negative bacteria A. baumannii LAC-4 (GCA_000786735.1)[38] and P. sp. Irchel 3E13..." to ".....negative bacteria A. baumannii LAC-4 (GCA_000786735.1)[38] and Pseudomonas sp. Irchel 3E13..."
11. • Introduce a line break between line 503 and 504.
12. • Page 14 line 543, please change "XbaI" to "XbaI"

Significance

This is an interesting manuscript that contributes to the knowledge of the legionaminic biosynthetic pathway and establish a glyco-profiling platform for the functional analysis of genes involved in pseudaminic (Pse) and legionaminic (Leg) acid biosynthetic pathways. The analysis of Leg patway allowed to identify a gene (legX) that can be used to distinguish Leg from Pse biosynthesis pathways, becoming a bioinformatic tool for the assignment and discrimination of these two pathways. Furthermore, a new class of FlmG protein glycosyltransferases, able to transfer Leg to the flagellin, has been identified and its analysis reveal two modular determinants that govern flagellin glycosyltransferase specificity: a glycosyltransferase domain that accepts either Leg or Pse, and a specialized flagellin-binding domain to identify the substrate.

I believe this should be "code talkers" as the link itself states.

All these should be under error

Also, all the links in the source and step descriptions are opening in the same tab. Ensure these links open in the new tab.

In the below sample error response, the description, source, step and reason indicates that the API failed due to authentication failure for incorrect OTP.

Author Response

Reviewer #1 (Public Review):

The software presented in this paper is well documented and represents a significant achievement that breaks new ground in terms of what is possible to render and explore in the web browser. This tool is essential for the exploration of SC2 data, but equally useful for the tree of life and other tree-like data sets.

Thank you for reviewing my work and for this generous assessment.

Reviewer #2 (Public Review):

This manuscript describes a web-based tool (Taxonium) for interactively visualizing large trees that can be annotated with metadata. Having worked on similar problems in the analysis and visualization of enormous SARS-CoV-2 data sets, I am quite impressed with the performance and "look and feel" of the Taxonium-powered cov2tree web interface, particularly its speed at rendering trees (or at least a subgraph of the tree).

Thank you for the kind words.

The manuscript is written well, although it uses some technical "web 2.0" terminology that may not be accessible to a general scientific readership, e.g., "protobuf" (presumably protocol buffer) and "autoscaling Kubernetes cluster". The latter is like referring to a piece of lab equipment, so the author should provide some sort of reference to the manufacturer, i.e., https://kubernetes.io/.

Thank you for flagging this. I have now replaced the colloquial "protobuf" with "protocol buffer". I have now provided a URL for Kubernetes. It is always difficult to judge how much to explain technical terms. I certainly agree that many people will be unfamiliar with, for instance, protocol buffers, but an explanation of what they are (which may not be particularly important for understanding Taxonium) can sometimes overshadow more important details. So my preference in that particular case is for an interested reader to research the unfamiliar term.

In other respects, the manuscript lacks some methodological details, such as exactly how the tree is "sparsified" to reduce the number of branches being displayed for a given range of coordinates.

This is an important point also raised by Reviewer 3. I have added a new section in the Materials and Methods which discusses this in some detail.

Some statements are inaccurate or not supported by current knowledge in the field. For instance, it is not true that the phylogeny "closely approximates" the transmission tree for RNA viruses.

I agree that this was an overly broad claim, and have softened it, now saying:

"The fundamental representation of a viral epidemic for genomic epidemiology is a phylogenetic tree, which approximates the transmission tree and can allow insights into the direction of migration of viral lineages."

Mutations are not associated with a "point in the phylogeny", but rather the branch that is associated with that internal node.

I have changed this as suggested.

A major limitation of displaying a single phylogenetic tree (albeit an enormous one) is that the uncertainty in reconstructing specific branches is not readily communicated to the user. This problem is exacerbated for large trees where the number of observations far exceeds the amount of data (alignment length). Hence, it would be very helpful to have some means of annotating the tree display with levels of uncertainty, e.g., "we actually have no idea if this is the correct subtree". DensiTree endeavours to do this by drawing a joint representation of a posterior sample of trees, but it would be onerous to map a user interface to this display. I'm raising this point as something for the developers to consider as a feature addition, and not a required revision for this manuscript.

I entirely agree with this point. I have added a sentence in the discussion:

"Even where sequences are accurate, phylogenetic topology is often uncertain, and finding ways to communicate this at scale, building on prior work [Densitree citation] would be valuable."

The authors make multiple claims of novelty - e.g., "[...] existing web-based tools [...] do not scale to the size of data sets now available for SARS-CoV-2" and "Taxonium is the only tool that readily displays the number of independent times a given mutation has occurred [...]" - that are not entirely accurate. For example, RASCL (https://observablehq.com/@aglucaci/rascl) allows users to annotate phylogenies to examine the repeated occurrence of specific mutations. Our own system, CoVizu, also enables users to visualize and explore the evolutionary relationships among millions of SARS-CoV-2 genomes, although it takes a very different approach from Taxonium. Taxonium is an excellent and innovative tool, and it should not be necessary to claim priority.

I agree that comparisons with existing tools are difficult and often provide a sense of unnecessary competition. I attempted to be quite careful in the specific section focused on comparison, but may have been less careful earlier on. The intent with this first sentence in the abstract was to provide a succinct description of the gap that Taxonium was developed to fill with "however, existing web-based tools for analysing and exploring phylogenies do not scale to the size of datasets now available for SARS-CoV-2". I have now removed the words "analysing and", focusing on the exploration of phylogenies. I think this new sentence is defensible in that valuable tools such as CoVizu intentionally do not explore a phylogeny directly but instead take a multi-level approach, and this new sentence better matches the comparisons in the paper. In the second sentence, I have removed the phrase "is the only tool that", which I agree adds little and may not be accurate, depending on one's interpretation of "readily". Thank you for these points.

Although the source code (largely JavaScript with some Python) is quite clean and has a consistent style, there is a surprising lack of documentation in the code. This makes me concerned about whether Taxonium can be a maintainable and extensible open-source project since this complex system has been almost entirely written by a single developer. For example, usher_to_taxonium.py has a single inline comment (a command-line example) and no docstring for the main function. JBrowsePanel.jsx has a single inline comment for 293 lines of code. There is some external documentation (e.g., DEVELOPMENT.md) that provides instructions for installing a development build, but it would be very helpful to extend this documentation to describe the relationships among the different files and their specific roles. Again, this is something for the developers to consider for future work and not the current manuscript.

This is an entirely fair comment. The version of Taxonium presented in the manuscript is "2.0", which is a new version built from scratch with considerably less technical debt than the version that preceded it. Its technical strengths are that (with the exception of the backend) it is relatively well-modularised into functional components. But the limitations that the reviewer notes with respect to commenting are entirely fair. What I would say is that in the time since this manuscript was submitted, several important features have been added by an external collaborator, Alex Kramer, most notably the Treenome Browser (https://www.biorxiv.org/content/10.1101/2022.09.28.509985v1). I hope that the ability of Alex to add these features with little need for support provides some evidence of Taxonium's extensibility. But I acknowledge there is room for improvement.

Reviewer #3 (Public Review):

The paper succinctly provides an overview of the current approaches to generating and displaying super-large phylogenies (>10,000 tips). The results presented here provide a comprehensive set of tools to address the display and exploration of such phylogenies. The tools are well-described and comprehensive, and additional online documentation is welcome.

The technical work to display such large datasets in a responsive fashion is impressive and this is aptly described in the paper. The author rightly decides that displaying large phylogenies is not simply a matter of rendering "more nodes", and so in my eyes, the major advancement is the approach used to downsample trees on-the-fly so that the number of nodes displayed at one time is manageable. This is detailed only briefly (Results section, 1st paragraph, 2 sentences). I would like to see more discussion about the details of this approach.

Thank you for this point, also raised by Reviewer 2. I have now added a lengthy section on this in the Materials and Methods, which I hope is helpful. The approach is not especially sophisticated, but it does the job and runs quickly.

Examples that came up while exploring the tool: the (well implemented) search functionality reports results from the entire tree (e.g. in Figure 4, the number of red circles is not a function of zoom level), how does this interact with a tree showing only a subset of nodes?

Yes, this is an important feature which I perhaps did not do justice to in the write-up. I have included in the new section in the Materials and Methods a paragraph discussing search results:

"In order to ensure that search results are always comprehensive, but at the same time to avoid overplotting, we take the following approach::

● Searches are performed across every single node on the tree to select a set of nodes that match the search. The total number of matches is displayed in the client.

● If fewer than 10,000 matches are detected, these are simply displayed in the client as circles

● If more than 10,000 matches are detected, the results are sparsified using the method above, and then displayed.

● Upon zooming or panning, the sparsification is repeated for the new bounding box."

How is the node order chosen with regards to "nodes that would be hidden by other nodes are excluded" and could this affect interpretations depending on the colouring used?

This perhaps was slightly sloppy language which did not directly describe the implementation. I have now rephrased this to "only nodes that overlap other nodes are excluded", as we don't in fact consider a notion of z-index when doing this. The way the sparsification works (now better described) means that the nodes excluded are determined essentially by position and I don't foresee this introducing particular biases, but this was an insightful point to raise.

Taxonium takes the approach of displaying all available data (sparsification of nodes notwithstanding). Biases in the generation of sequences, especially geographical, will therefore be present (especially so in the two main datasets discussed here - SARS-CoV-2 and monkeypox). This caveat should be made explicit.

This is certainly true. I have added this new paragraph in the Discussion:

"A further challenge is the vastly different densities of sampling in different geographic regions. Because Cov2Tree does not downsample sequences from countries which are able to sequence a greater proportion of their cases, the number of tips on a tree is not indicative of the size of an outbreak and in some cases even inferences of the directionality of migration may be confounded. There would be value in the development of techniques that allow visual normalisation of trees for sampling biases, which might allow for less biased phylogenetic representations without downsampling."

Has the author considered choosing which nodes to exclude for sparsified trees in such a way as to minimise known sampling biases?

The last sentence of the new paragraph above alludes to a sort-of-similar approach. I hadn't directly considered the approach the reviewer suggests. It is an interesting idea. The downsampling algorithm has to be very computationally inexpensive but it would be interesting to explore ways to do this. I am tracking this in https://github.com/theosanderson/taxonium/issues/437.

Interoperability between different software tools is discussed in a technical sense but not as it pertains to discovering the questions to ask of the data. As an example, spotting the specific mutations shown in figure 3 + 4 is not feasible by checking every position iteratively; instead, the ability to have mutations flagged elsewhere and then seamlessly explore them in Taxonium is a much more powerful workflow. This kind of interoperability (which Taxonium supports) enhances the claim of "providing insights into the evolution of the virus".

Thank you for flagging this point -- I am very excited by the growing ecosystem of interoperable tools. You are absolutely right that most of the insights Taxonium can bring into evolution rely also on this broader ecosystem. I have added a florid sentence in the concluding paragraph: "It forms part of an ecosystem of open-source tools that together turn an avalanche of sequencing data into actionable insights into ongoing evolution."

The prosaic reason I don't discuss Taxonium's interoperability features in more detail in this manuscript is that it aims to describe the version of Taxonium I initially developed, and these features were developed collaboratively by a broader group later on (and after deposition of this preprint).

Taxonium has been a fantastic resource for the analysis of SARS-CoV-2 and this paper fluently presents the tool in the context of the wider ecosystem of bioinformatic tools in use today, with the interoperability of the different pieces being a welcome direction.

Reviewer #2 (Public Review):

This manuscript describes a web-based tool (Taxonium) for interactively visualizing large trees that can be annotated with metadata. Having worked on similar problems in the analysis and visualization of enormous SARS-CoV-2 data sets, I am quite impressed with the performance and "look and feel" of the Taxonium-powered cov2tree web interface, particularly its speed at rendering trees (or at least a subgraph of the tree).

The manuscript is written well, although it uses some technical "web 2.0" terminology that may not be accessible to a general scientific readership, e.g., "protobuf" (presumably protocol buffer) and "autoscaling Kubernetes cluster". The latter is like referring to a piece of lab equipment, so the author should provide some sort of reference to the manufacturer, i.e., https://kubernetes.io/. In other respects, the manuscript lacks some methodological details, such as exactly how the tree is "sparsified" to reduce the number of branches being displayed for a given range of coordinates. Some statements are inaccurate or not supported by current knowledge in the field. For instance, it is not true that the phylogeny "closely approximates" the transmission tree for RNA viruses. Mutations are not associated with a "point in the phylogeny", but rather the branch that is associated with that internal node.

A major limitation of displaying a single phylogenetic tree (albeit an enormous one) is that the uncertainty in reconstructing specific branches is not readily communicated to the user. This problem is exacerbated for large trees where the number of observations far exceeds the amount of data (alignment length). Hence, it would be very helpful to have some means of annotating the tree display with levels of uncertainty, e.g., "we actually have no idea if this is the correct subtree". DensiTree endeavours to do this by drawing a joint representation of a posterior sample of trees, but it would be onerous to map a user interface to this display. I'm raising this point as something for the developers to consider as a feature addition, and not a required revision for this manuscript.

The authors make multiple claims of novelty - e.g., "[...] existing web-based tools [...] do not scale to the size of data sets now available for SARS-CoV-2" and "Taxonium is the only tool that readily displays the number of independent times a given mutation has occurred [...]" - that are not entirely accurate. For example, RASCL (https://observablehq.com/@aglucaci/rascl) allows users to annotate phylogenies to examine the repeated occurrence of specific mutations.<br /> Our own system, CoVizu, also enables users to visualize and explore the evolutionary relationships among millions of SARS-CoV-2 genomes, although it takes a very different approach from Taxonium. Taxonium is an excellent and innovative tool, and it should not be necessary to claim priority.

Although the source code (largely JavaScript with some Python) is quite clean and has a consistent style, there is a surprising lack of documentation in the code. This makes me concerned about whether Taxonium can be a maintainable and extensible open-source project since this complex system has been almost entirely written by a single developer. For example, `usher_to_taxonium.py` has a single inline comment (a command-line example) and no docstring for the main function. `JBrowsePanel.jsx` has a single inline comment for 293 lines of code. There is some external documentation (e.g., `DEVELOPMENT.md`) that provides instructions for installing a development build, but it would be very helpful to extend this documentation to describe the relationships among the different files and their specific roles. Again, this is something for the developers to consider for future work and not the current manuscript.

no data or code included

!- meme : intentional software as "immortal form" - existing legacy articulations are assets - to which new servicdes can be tied to - composed of, re-mix re-use, re-purpose

!- meme : intentional software - beyond programming - beyond ontology

• point to : long tail software

• arbitrary type calculus
• meta-work
• domain specific compilation knowledge?
• comments should be expressible

• as part of the machine processable part

• not literal programming

• where narraitve is interpersed and organizing
• code blocks

!- response : - but articulate software as a conversations and mutual learning

• no incentive to go outside the framework
• descending to machine code

I don't see any issues. Great job, Ravi.

Looks great!

Excusing criminality in a matter of deference to foreign nationals who are unlawfully present in the United States is perverse. The immigration laws have many provisions by design to ensure that foreign nationals who violate the laws of the United States in certain ways are not allowed to remain and harm the safety of Americans.

We can relate this to one of the first project we completed in this class. We had to retrieve API Keys from Twitter Developer to insert them into Python.

The code in the CSS looks great and well organized. I find the added in comments in the "/*" brackets is very help to show what the code is for.

The CSS on assignment-c-dlu-39 pancakes looks great. The code is looking correct and well organized.

The HTML on assignment-c-dlu-39 pancakes looks great. The code looks correct and it's organized well.

As this is the root element you can put this code at the top of your css sheet

answering using source code

Lady Harris genuinely tries to form relationships with Indian women and appeal to activities, like theater, that they would enjoy, instead of forcing them to participate in western forms of education

Very clean and organized font organization. I found it quite easy to quickly get to the font or style I'm looking for. I will try to emulate this in my future code organization.

there is a big opportunity cost to unassisted learning to code: many people will drop-out (in frustration). Again this is how the IndieWeb community quickly drives people away too. the skill level acquired this way is limited (due to absence of Vgl deliberate practice?) the acquired skills do not make following activities more efficient. (Vgl how I usually struggle with the same basic coding issues, no matter how many scripts I try to write).

Still very common to observe as example of such 'coder's ethics'. E.g. IndieWeb community discussion is mostly based on figuring things out yourself and creating one's own code (self-dogfooding), with only generic and hand-wavy assistance most of the time.

The way you learned programming shapes assumptions about how to learn programming.

The fact I self-learned through trial and error without supervision in the early and mid 80s, is common, but not a model for K transfer, and not useful "coder's ethic"

freeze makes the requirements.txt like a package.json.

(Author:: [[neurino on Stack Overflow]]) python from urllib.request import pathname2url pathname2url('dir/foo, bar.mp3') ∎

sometimes, I get coupons code from McDonald give me special coupon codes, which my other friend don't get, for example, I get a 50% off for a ice cream cone, but my friend just got 30% off, we don't know why and what makes the difference, but maybe McDonald's think I can only afford 50% for the cone, they want me to spend money on it, so they give me a higher discount. I am not sure if this can be considered as "fair", but I think this coupons are based on personal income, and asset.

Valentine, Greta, and Kate Barron. 2022. “An Examination of Academic Library Privacy Policy Compliance With Professional Guidelines”. Evidence Based Library and Information Practice 17 (3):77-96. https://doi.org/10.18438/eblip30122.

Abstract

Objective – The tension between upholding privacy as a professional value and the ubiquity of collecting patrons’ data to provide online services is now common in libraries. Privacy policies that explain how the library collects and uses patron records are one way libraries can provide transparency around this issue. This study examines 78 policies collected from the public websites of U.S. Association of Research Libraries’ (ARL) members and examines these policies for compliance with American Library Association (ALA) guidelines on privacy policy content. This overview can provide library policy makers with a sense of trends in the privacy policies of research-intensive academic libraries, and a sense of the gaps where current policies (and guidelines) may not adequately address current privacy concerns.

Methods – Content analysis was applied to analyze all privacy policies. A deductive codebook based on ALA privacy policy guidelines was first used to code all policies. The authors used consensus coding to arrive at agreement about where codes were present. An inductive codebook was then developed to address themes present in the text that remained uncoded after initial deductive coding.

Results – Deductive coding indicated low policy compliance with ALA guidelines. None of the 78 policies contained all 20 codes derived from the guidelines, and only 6% contained more than half. No individual policy contained more than 75% of the content recommended by ALA. Inductive coding revealed themes that expanded on the ALA guidelines or addressed emerging privacy concerns such as library-initiated data collection and sharing patron data with institutional partners. No single inductive code appeared in more than 63% of policies.

Conclusion – Academic library privacy policies appear to be evolving to address emerging concerns such as library-initiated data collection, invisible data collection via vendor platforms, and data sharing with institutional partners. However, this study indicates that most libraries do not provide patrons with a policy that comprehensively addresses how patrons’ data are obtained, used, and shared by the library.

looks great

This is linking to the "putting it all together" (pose.ipynb) example. Is it meant to link to the "Point cloud registration with known correspondences" in the icp notebook?

indent

For Code readability i would have put a blank line above this section just to be able to discern.

use this for explaining values

use when talking about the values

for part 1 about what code of ethics are and why they are important

对象、函数、解释器，掌握各个概念之间的关系和它们在代码中的角色，对编码能力非常重要。

Meiner Meinung nach ein sehr interessanter Zug von Google. Sie verzichten lieber auf die Weiterentwicklung ihrer Software, die die Konkurrenz potentiell einsehen könnte, um die eigene Machtstellung zu sichern und auszubauen.

.code (Author:: [[thmsnhl on Stack Overflow]])

Need some code in the footer 1.Change the background colour 2.Add the declaration padding: 40px; 3.Change the colour of the links in the footer

redundant, in longer pages of code it will slow down your page. Just use one <link>

This instance, among others, seems like excessive ID use. Consider using descendant selectors like .card-three h1 {} to make your code more concise.

.c1

Yeah, I've noticed this. The argument is that if a new type is added you may need to update many switch statements to support the new type.

Hello SENG.

This is an annotation added on top of this web page. This is to illustrate how we can add notes on top of documents on the web using Web Standards.

It's also my way of using code to illustrate ideas.

this doesn't link to anything. you need to make it <a href="contact.html"> This is also duplicate code.

Anecdotes of self-censorship to avoid algorithmic censors

A few paragraphs later in the article there is a story from a high school literature teacher that wont use the word "death" lest it might "stunt his reach."

I agree with Abbey regarding the P element but leave in the div, and add h3 along with the p. I had some help from Claire , Use the div class for both p and h3 right after the image code.

for the border declarations it can be shortened by using shorthand: border-top: 3px solid hsl(...); this gives all the info in one line of code

idk how to use this hypothesis thing but here we are

You looks absolutely good to me, I learned some new things from your code. Great work

Good job using comments to source code!

I'm not sure how to be constructive or critical of your code, it looks better than mine lol

Do you have any footer list items? I couldn't see any, if none are used I would remove this extra code.

same with this comment, place next to (or otherwise near) the code it refers to.

I would place this comment next to the code you're referencing. Right now it's hard to tell what this goes to.

for this I might have just made the phone number the hyperlink, the "click here" throws off the flow when reading it, but the code itself is great!

don't-understand the code or mbr called boot loader

Linked a source, nice!

you can do some changes in card four according to instructions ,moreover your code is good

Html code looks well formatted

The Ghost Dance was created by American Indians, in response to the unwanted presence/intrusions of white settlers on their territory. As more and more settlers pushed further west, Native populations not only lost the land they had occupied for hundreds of years but the resources they relied upon for just as long. White settlers, in this context, can reasonably be seen as an invasive species to the American landscape. However, rather than out-competing Native populations, colonial settlers dwindled their communities with the disease, forced migrations, and environmental degradation. These circumstances were so drastic to American Indians that they came up with a technique or code, in the Ghost Dance, to help bring about an end to the world they have been forced into. In doing so, they would unlock a new existence free of white settlers who are continually encroaching upon and inflicting themselves on your life. With all this in mind, I would be inclined to agree with Sclonit's assessment of the Ghost Dance from a technological standpoint. It is an organized and systematic practice meant to bring about a specific desired result or experience, in this case, a new world.

Purpose code, not PAN details

Automatizar páginas web.

Black student have to code switch to be accepted academically, this promotes linguistic supremacy because it is teaching all student that what is accepted academically and professionally is not black

Code-switching is something that occurs with bilingualism, and this point is demanding that teachers acknowledge and use the connections between language rather than a strict structure of only white mainstream English as that is a form of injustice.

Students should not have to code switch

This is a very good method that I can smoothly complete my work. It let me know how to find bot passwords. And I can learn a skill.

I believe that in today society, our public persona have gotten very similar to one another, we as a society dislike the idea of being different, and in some country it is frown upon to stand out, like in the case of Japan and many Asian country. people are have different personality depends who they are with, like how you would behave around your parents versus how you are with your friends or significant other.

A person may have different persona for different situations. Though the persona may differ from who he really is, but it doesn't necessarily mean that the person is inauthentic. Sometimes, a modification of one's behavior is needed just like the following examples indicate. A person's persona may subconsciously change with the surrounding environment.

• https://github.com/innixma/fall-automl-2022-autogluon
• https://github.com/gidler/autogluon-tutorials

explanation of the relation to they hypothesis and the conclusion drawn from the data seen/measured

In to which file should I put this code?

⬑ what khinsen calls software collapse

⬑ "The Code Rot Problem"

This page has a neat effect, first apparent with this example, where a blur effect is used on most of the text in the code block, except for lines 11–13 which are shown in sharp focus. (You can mouse over the code block to eliminate the blur effect.)

.torchlight.has-focus-lines .line:not(.line-focus) {
        transition: filter 0.35s, opacity 0.35s;
        filter: blur(.095rem);
        opacity: .65;
}

Each line is dumped into a div and the line-focus class set on those which are supposed to be unblurred.

(For ordinary code blocks without any blur/focus effect, the has-focus-class line is simply not used.)

I think this is amazing to do but also could have some draw backs. It would need to have some sort of security measure as I think some text in any language, religion, culture is considered sacred and should be treated with respect but obviously the world does not work as it should all the time. In my opinion the collection should have two layers to it, one which would be open to general public and a second that can only be accessed through a certain code (which could be through application/online quiz/ etc..) that only allows people that actually care enough to learn to get access what is considered to more "sacred" important files. (although all files would be important)

Please add the sample code. I think you will need to add two. One for when payout is made to card and one when payout is made to token

the mode is card as per the sample code. it is not mention here in the possible values. please check this once

Similar to how game developers need to code a script

Section name too long. Just "Examples" is enough

Clean code that is easy to read, no needed edits just some suggestions that you can ignore.

Google has an embedded maps you can add, google maps>click on the share button>click on the embed a map tab and copy the code.

Similar comments - I would use unordered lists and paragraph elements where necessary!

Also, I found there is an < address > tag you could use. Learn more: https://www.w3schools.com/tags/tag_address.asp

https://github.com/ledell/automl-fall-school-2022

See GigaBlog forr more http://gigasciencejournal.com/blog/frictionless-data-interactive-figures/

See more in GiigaBlog on how these were created http://gigasciencejournal.com/blog/frictionless-data-interactive-figures/

I suppose I get it now, there is a subtle underlying theme in their dynamic of an algorithm, the algorithm of them, like when George talked about a deviation in their code or something

MMMM THIS LANGUAGE it feels like George

Based on lots of tagging of endogenous loci using a similar plasmid based delivery system in C. elegans, I would think that you should be able to get away with moving the insertion site to the C-terminus (in this case) or N-terminus (in other cases) - it would be great to test this out - but we (and others) have found that if you re-code the homology arm in your repair plasmid between the cut site and insertion point, you can get KIs when you push the distance between the cut site and the insertion site, avoiding having to insert your FP into the gene. Is there a locus of interest with some different sgRNAs at the C- or N-terminus that you could experimentally try this out with (like one that would split the insertion site, one that would be ~10-20 bp away, and one that would be ~50-60bp away)?

Would they even know what they were looking at?

potential government source.

This code is well written and find it inspiring. Great work.

Nothing in your code stands out to me as anything I wouldn't do :)

I'm so excited your code is open source! Your website is so beautiful I assumed the code would be hidden. Very excited to explore everything there.