Sniffbot: A Bio-Hybrid Robot for Odor Localization and Discrimination.(A) The antenna of the desert locust is removed and placed in the antenna holder within Sniffbot, which employs active sensing to sample the surrounding air for odor cues.
It looks like figure 1A has been transformed and compressed, hard to understand what is going on there. I think figure1C and figure 1D were accidentally added there?
Upon final publication, we will release a complete open build package designed to enable full replication and modification of the OLH platform. This package will include mechanical CAD files and 3D-printable part models, a detailed bill of materials with catalog numbers and supplier information, wiring diagrams and I/O mapping tables for all electrical and control components, step-by-step assembly and calibration documentation, and the full source code for hardware control, experiment scheduling, and data analysis. We are finalizing this documentation for public release and it is not yet posted in full due to the scope of the materials involved. For status updates or early access to in-progress documentation, contact emma.chory{at}duke.edu.
Thank you for sharing the Open Liquid Handler, it's really inspiring to see a platform like this being built from scratch and implemented in the lab! Whenever the full repository is released, it would be great to see some more experiment details on the high throughput turbidostat. Specifically, the organisms being grown, the media types, plate types, how exactly the shaker was implemented, and OD data for your experiments. I look forward to learning more about the OLH when the build package is released!
Supplementary InformationMicrofluidic device fabricationDetermination of the gap height of the microfluidic chips
The supplemental information is not available
Further, , utilizing the 3D/4D AI perception through camera feeds, we demonstrate a proof-of-concept cobot module for LabOS, as in (Fig. 3g, Supp. Video 2). This module enablesscientists to run time-consuming and/or repeated portions of the protocol using an adaptive, andspatially aware robotic arm. We demonstrate example protocols loaded in LabOS, such asvortexing, 96-well plate operations, tube handling on incubator/shaker, can be handed off to thecobot to reduce the load on scientists, where agentic and embodied AI with spatial intelligenceprovide a seamless human-AI-robotics collaboration experience (Supp. Video 3).
I really like the idea of the xr glasses collecting data that can then be used to inform the world model for the co-bot, fascinating approach!
userssimply load or drag our BOTany protocols and updated CSV files onto Opentrons App toexecute the procedures, thereby providing a direct route to laboratory automation.
I think taking advantage of Opentrons csv parameter abilities to modify scripts is a great approach! I'd be curious to know if you have come across any edge cases when users are modifying tables in excel and how you have approached them? Things like adding spaces, capitalizations, typos etc
(Supplementary Fig. S3)
I think in part A of this figure the illustration describing the action in between plates during the manual steps is not immediately clear. I think it is supposed to represent a multichannel pipette?
LabscriptAI achieved 89.1% overall success in generating simulation-passing scripts, outperforming all baselines (Table 1)
It would be interesting to know what percentage of these scripts you were able to run on the opentrons, I've seen scripts that pass simulation but then immediately throw errors when uploaded to the robot
Performance was compared against direct LLM implementations (GPT-5, Claude 4, DeepSeek V3.2, Gemini-2.5 Pro) and OpentronsAI, a commercial LLM–based solution (Table 1).
I'd be interested to see this comparison done with other agentic approaches like Claude Code or Cursor if they were given the same kind of context upfront
The 233black microplate exhibits substantially reduced background interference, making it a 234superior choice for Raman screening applications.
Any hypothesis for why this might be? It would also be helpful to include the name and part number for both plate types
This design enables firm attachment of the plate frame to the build plate. Figure 4:B shows the microplate mounted in the frame and affixed to the build plate.
I really like this idea for mounting 96 well plates!
Positioning error.The variation in PLA spectrum due to the mispositioning caused by different speeds in A: Z movements B: XY movements
Descriptions missing for C and D
The chip was fabricated by soft lithographyas previously described
Where has this been previously described? A link to a citation here would be helpful
we systematicallyevaluated seven laser intensities (0.01%, 0.1%, 1%, 3.2%, 5%, 10% and 25%) with a 532 nmexcitation source, 100× objective, and 800-3200 cm⁻¹ spectral rang
You mention the relative laser intensities, it would be helpful to provide the laser power as well.
However, when the objective is to express or screen proteins in high-throughput, the requirement for single colonies can be relaxed, and non isogenic cultures can be used.
It was great to read how you were able achieve this on the OT2. I'm wondering, if cultures can be non isogenic, wouldn't it be more straightforward to transfer transformed cultures to selective liquid media plates instead of dealing with the complexities of calibrating for agar plates? I'd love to hear your thoughts, thanks!
As this tool simplifies the programming process, it also allows researchers with limitedcoding experience to leverage the full capabilities of the OT-2
I really like this approach to making structured and reproducible Opentrons protocols without generating entire new scripts each time. I'm curious how LLMs might fit in upstream of this, could you describe your protocol in plain language, have the LLM generate the csv files, then run protocol generation script? Could the LLM just generate the protocol at this point?
Through the analysis of these data, the neural effects of chemicals or drugs296can be evaluated and the scientific basis for the safety assessment can be provided.
Really fascinating method for culturing, screening and imaging all on one platform. Using pillars to 'cage' the c. elegens while allowing liquid to pass through is a great idea. I'd be curious if you have thought of any ways this design could be adapted to sort c. elegens into collection chambers based on their phenotype?
The narrow geometry of 96 well plate wells allows us toavoid the complexity of a variable tilt mechanism as commonlyseen in automated cell culture systems [5, 6]
I really like this implementation of tilting the plate with a custom holder instead of a variable tilt mechanism, will consider using something like this for our opentrons!
The Petri dishes with the cultures were then irradiated with UV light for 5 minutes in119a sterile hood.
It would be helpful to know what the power output of the light is, what wavelength it emits, and how far the light source is from the petri dish.
To counteract this, we retroreflected the illumination line as shown in Figure 1C & D, such that the flowing sample was illuminated (pushed) from both sides.
Instead of adding the second beam to counteract the force of the light sheet, is it possible to add a 3D sheath flow that pushes cells toward the bottom of the channel? This paper explains this concept https://doi.org/10.1063/5.0033291
We recently introduced a new method of vat-based 3D printing that we refer to as InjectionContinuous Liquid Interface Production (iCLIP).
iCLIP looks like a really promising tech for making 3D printed microfluidics, it was great to learn more about it through your paper. I wonder how long you could have resin 3D printed structures like these in your skin without negative side effects? I could imagine devices like these being built into wearables that can inject drugs continuously over longer periods of time
Looking ahead, future efforts will focus on developing lower-cost hardware solutions to further democratize access to microfabrication techniques. Digital light processing (DLP) printers contain much of the necessary hardware (UV lamp, driver board, DMD)
I'm really looking forward to seeing how this goes and would be interested in building one myself!
The use of 3D printing resin as a substitute for SU-8 eliminates time-sensitive baking steps and reduces the need for extensive glass slide cleaning.
I think its great to be able to use 3D printer resin and still get high quality structures. I'd be curious to see a comparison of devices made with SU-8 vs 3D printer resin using this maskless approach, are there still benefits to SU-8 that make investing in the proper safety equipment worth it? It would also be interesting to see if there is still any benefit to using silicon over glass
Clear
Might be worth trying a darker opaque resin, this paper showed a reduction in surface roughness when using phrozen High Temp 3D Printer Resin [TR250LV]. https://doi.org/10.1038/s41378-023-00607-y
However it remains low enough to guarantee optical transparencyfor cell observation by confocal microscopy, and our molds exhibit similar roughness to what waspreviously shown for similar mold printing techniques [23].preprint (which was not certified by peer review) is the author/funder. All rights reserved. No reuse allowed without permission.The copyright holder for thisthis version posted January 29, 2025.;https://doi.org/10.1101/2025.01.29.632980doi:bioRxiv preprint
I'm curious at what orientation you printed these molds? This paper showed a reduction in surface roughness when tilting multiple axis of the model. https://doi.org/10.1038/s41378-023-00607-y . Not sure if this same technique will translate between DLP and SLA though
Thickness measurement of the PDMS layer under the BOTTOM channeldepending on molding techniques
I think it would be helpful to label Figure 3E with the molding techniques to get a better sense of what is actually happening to the PDMS during the curing step
A) 3D reconstitution of the printed patterns imaged witha surfaced microscope. B) close up photo of one of the printed patterns for a TOP mold.
I think the description for A and B in Figure 2 were swapped
salinized
should be 'silanized' with an 'i'
Additionally, Reynolds number (Re) estimations suggested that the maximum Re inside the chamber was less than 1, corresponding to laminar flow
It might be helpful to walk through what dimensions went into this Reynolds number estimation.
Fully cured PDMS parts were separated from the molds using a craft knife.
What were the conditions for the PDMS curing?
Supplemental Figure 1 provides images of standard USAF resolution targets for the objectives and configurations used in this investigation.
It's difficult to understand what exactly the distance vs gray scale chart is referring to underneath each of the USAF targets. A short explanation here would be helpful.
The X,Y-resolution (effective pixel resolution) was similar between the two printers, at 30 and 40 μm for the 405- and 385-nm printers, respectively. In this resin, the crosslinking reaction was more efficient with the 385 nm light source, enabling reduced light dosage (shorter times and lower intensities; Table S1), which assisted in diminishing bleed-through light allowing uncured resin to drain more easily from channels. Consistent with this, channels were printable at sizes ~0.2 mm smaller with the 385 nm printer (Fig. 2C) versus the 405 nm source (Fig. 2B).
Thank you for including this information, I have tried similar to make similar microfluidic channels of different dimensions, but did not consider the differences that light source wavelength would have on different resin types.
MER2-1220-32U3M, Daheng Imaging
It was great reading about an open-sourced imaging platform being used as a low cost method of disease detection. I'm curious how this system works with other cameras and what trade-offs there are between this camera and others you may have bought. I've been using FLIR cameras for open-source affordable imaging projects, but I am not sure that is the best option.
The results prove that 1-minute mechanical maceration using such a small handheld device can perform almost equivalently in sample lysis to manual grinding using a mortar and pestle.
Really great reading about this device! Has anyone tried using it for nucleus extractions of algae? We have used a modified version of this protocol where we cryo-grind Chlamydomonas with a mortar and pestle, but found it to be laborious and inconsistent. I'd imagine you'd need to make significant modifications to cool the system, though that might be beneficial due to the heat that is generated from the motor
The low-cost imaging platforms presented here provide an opportunity for labs to introduce phenotyping equipment into their research toolkit, and thus increase the efficiency, reproducibility, and thoroughness of their measurements.
I really like this approach to plant imaging, I'm curious what you think about a system like this that allows the camera to move between positions to take time lapse images of even more samples?
The low-cost imaging platforms presented here provide an opportunity for labs to introduce phenotyping equipment into their research toolkit, and thus increase the efficiency, reproducibility, and thoroughness of their measurements.
I really like this approach to plant imaging, I'm curious what you think about a system like this that allows the camera to move between positions to take time lapse images of even more samples?
Schematic of experimental approach: New Zealand white rabbits are implanted with a chronic32-channel ECoG grid over visual cortex, and visually-evoked potentials are recorded as highcontrast stimuli are presented using a monitor.
I've really enjoyed reading this paper, learned a lot about optigenetics and ocular implanted devices. I'm curious how you plan to validate the effectiveness of these devices in animals and what kinds of assays would make the most sense.
laboratories.
I really enjoyed reading about this project and am curious about implementing something like this for our lab! It would be great to see a video of it in action. It would also be great to know how quickly COPICK can pick colonies from a plate vs a person in the lab?
At present time (early 2023), there is still a lack of open-source tools on the web to label and create datasets of images using a panoptic segmentation format in a straightforward fashion.
Has this changed in the last year? Would love to know where advancements are being made
using a reflex camera with a macro objective (Nikon D60 with AF-S Micro NIKKOR 60 mm f/2.8G ED lens)
I'm curious if there are any drawbacks to training the model using a different camera than the one that is implemented on the OT-2?