A standardized, realistic phantom dataset consisting of ground-truth annotations for 6 diverse complexes serves as a community resource for benchmarking cryo-ET algorithms. www.nature.com/articles/s41...
Images, AI, ML, and ALife. Former: @hampshirecolg @brandeisuniversity.bsky.social @harvardmed.bsky.social @uidaho.bsky.social @hhmijanelia.bsky.social @mdc-berlin.bsky.social @ORNL. Now @chanzuckerberg.bsky.social. Opinions are mine.
1,107 followers 2,301 following 40 posts
view profile on Bluesky
A standardized, realistic phantom dataset consisting of ground-truth annotations for 6 diverse complexes serves as a community resource for benchmarking cryo-ET algorithms. www.nature.com/articles/s41...
Caroline Hemphill also wrote up a great blog post about the competition: www.czbiohub.org/life-science...
Better ML for cryo-ET starts with better benchmarks. We built a phantom cryo-ET dataset (~500 tomograms) + hosted a Kaggle challenge. The result: community models beat expert tools. Read more in the @natmethods.nature.com article that just came out: 🔗 doi.org/10.1038/s415...
Is that an anvil?
It's hard to express how much gratitude I have to everyone involved in the journey, both mine and @napari.org's, that has led to this point. #EurekaPrizes A small attempt below: 👇
🚀The napari team is excited to announce our brand new development roadmap — a community-informed vision for the future of napari! 🔍Read the blog to learn about our process and priorities: napari.org/island-dispa... 🗺️OR check out the roadmap itself on napari.org: napari.org/stable/roadm...
If you post the frames in a thread then we can try scrolling really fast
when you discover new tools and ideas via biorixv, sometimes you don't even notice when it is published, some great stuff here : www.nature.com/articles/s41...
bsky.app/profile/wais...
There is a @napari.org hackathon at GloBIAS! Be there or be square (or whatever layer type you'd like)
TFW you discover the internet in the conference venue isn't stable enough for a Google Slides presentation with videos 😅 luckily with >24hr warning
And it's finally here! napari 0.6.0 is out on PyPI and conda-forge! 🥳 We've had a few threads about this release already but let's recap with some highlights: 🧵
We have a suite for studying cell polarity in Python polarityjam.readthedocs.io feel free to post an issue if you need any more functionality
We have updated our #MemBrain v2 preprint with a lot more details about the MemBrain-pick and MemBrain-stats modules, as well as some application examples! Stay tuned for the upcoming thread by lead author @lorenzlamm.bsky.social! 🧠🧵 #CryoET #TeamTomo www.biorxiv.org/content/10.1...
Yo #TeamTomo, check out our updated #MemBrain v2 preprint. And better yet, give it a whirl on your #CryoET membranes! Please send us your feedback! 🧪🧶🧬🔬
"no no, I said turn up the economy"
Check this out!!! The results from our CryoET particle picking competition are available, the full test data (public and private datasets) is available, and you can dive right into the data on the CZ CryoET Data Portal! Enjoy and please let us know what you think :)
Awesome! So glad to see my former RL professor Andy Barto get this recognition (and Rich is too!)
Last night we were doing a family dance party... My smartwatch gave a notification... You've been running for 5 minutes... I feel judged.
Amazing new #Python #PEP to improve optional dependencies handling: https://peps.python.org/pep-0771/ The PEP is very thorough and well-written, and it perfectly uses @napari as a motivating example. We were not involved but we fully endorse it! 😃
Hey #teamtomo, if you often find yourself using the excellent membrain-seg from @lorenzlamm.bsky.social et al., you might find my napari plugin "napari-segselect" useful. Let's say your tomogram contains the edges of two bacterial cells, each with a membrane and cell wall:
I just want to take the opportunity to remind everyone that the Hamster Dance still exists. You can even find 10 hour long videos of it. YW
There's a Korean soap opera on Netflix (When the Stars Gossip) which has *Drosophila courtship in space* as a major plot point (Ep 2). They have Michael Reiser & lab's flight arenas (you can see Janelia & MIchael's initials on the panels). Great way to learn about Drosophila courtship!
TL;DR installers are really challenging to make. The team made some tooling for @napari.fosstodon.org.ap.brid.gy that was so fantastic that it is spreading across the scientific python ecosystem!
🚨Looking for a new opportunity at the intersection of #AI and #bioimaging? Want to work with an awesome team? We have multiple open research scientist positions @cziscience.bsky.social. Please RT! boards.greenhouse.io/chanzuckerbe... boards.greenhouse.io/chanzuckerbe...
We just released and uploaded a new version of *Mastodon*, the large-scale tracking and track-editing framework for large, multi-view images. Mastodon is available via Fiji, via a dedicated update site. mastodon.readthedocs.io/en/latest/do... Below is a thread that survey some of the novelties ⏬️
It was the user's fault for not signing up sooner.
🤣
is the strategy for getting gemini users to have pop up windows in google apps?
AI startup recruiting emails these days: "Our company coming out of stealth mode w new capital. We need a founding engineer for our startup that is focused on developing foundation models of founding engineers. Our plan for monetization is replacing founding engineers in post-stealth mode companies"
For some reason it feels very judgy when my smart watch says "Good job being active again" when I was sitting at my desk scratching my head.
I feel your pain. I had to scavenge some emergency J48s a couple of weeks ago.
Perfect, you've got it! github.com/oanegros/Mic...
This is beautiful!!! Do you happen to have some scripts for using Microscopy Nodes w/ bpy in python?
This should work for running it on your own machine if you have `uv` installed: ``` uv run atrium.kyleharrington.com/visualizatio... s3://janelia-cosem-datasets/jrc_mus-liver/jrc_mus-liver.zarr/recon-1/em/fibsem-uint8 --colormap gray ``` This is a video from OpenOrganelle of a Mouse Kidney
For the @napari.fosstodon.org.ap.brid.gy and large image data out there, I made some updates to the tiled loading zarr demo. It still has limitations (scale isn't respected, etc.), but this is an easy way to run it on your own code. github.com/napari/napar... More in 🧵👇
This week I tested pixel based machine learning using a Napari plugin written by @kephale.bsky.social , and posted a video of my experiments (interestingly Herbie G. may have beat both DL and ML with an ImageJ Macro on one of the test images) youtu.be/9rnNDkvVUyY
Yes! I'll write some new code to wrap up the last thing for the year!!!
halp
First NGFF Mesh call in 42 mins: github.com/ome/ngff/iss...
2) If you are interested in more advanced Machine Learning pixel classification, check out ConvPaint (github.com/guiwitz/napa...) and @kephale.bsky.social 's Interactive Napari Machine Learning tutorial (github.com/kephale/napa...).
Happy to be in sunny☀️San Diego for ASCB Cell Bio! Tomorrow I will present #Ultrack and #inTRACKtive (poster B513) at 11:15 am. Swing by the CZI+@czbiohub.bsky.social booth (#119) for live demos of these tools + #napari & #zebrahub 🐠 Looking forward to connect!🤩🧪🔬 #CellBio2024 #ASCB24 #celltracking
very glad to hear that. @craigreynolds.bsky.social is super cool! we've had exchanges about coevolving camouflage over the years, and Craig always keeps coming up with new ideas. of course boids were a great finding, but Craig's tag paper helps set the stage for the work on coevolved camouflage.
There is a "Women in AI" pack: bsky.app/start/did:pl... I'd also be happy to be added to your pack. I do ML for Systems research along with bioplausible networks.
The preview of the @chanzuckerberg.bsky.social Virtual Cell Platform just dropped and there is a nice post about behind the scenes of the CZII CryoET Object Identification challenge: Read more: virtualcellmodels.cziscience.com/benchmark/cr... Participate: cryoetdataportal.czscience.com/competition
For the #ML and #CryoET folks out there: Are folks on here submitting to the CZII CryoET Object Identification competition: www.kaggle.com/competitions... ? The leaderboard is really interesting, and the forums are full of insights. How many of the latest picking models have been submitted?
Of course it is critical to note, that the idea of using GP for prey pattern generation was started by @craigreynolds.bsky.social a few years earlier. However, the expectation in 2014 was that coevolution would lead to mediocre stable states in predator-prey games, and we disproved that by example.
This work led to awesome camouflage, but I do believe that the power of this approach comes from the coevolutionary/adversairal dynamics. With a visual attention system + genetic programming of prey patterns, we consistently achieved camouflage patterns. blogs.biomedcentral.com/bmcseriesblo...
Good point, using the micromamba dir in that example is a bit of a red herring. I think using a ~/.envs should be fine.
and i guess i'll just finish this by saying that if you want to manage more than just python packages (like cuda as you mentioned), which is the main reason i've used conda, that is when you want to start using pixi instead of just uv
Like, this is terrible but I can do it: uv venv ~/micromamba/envs/test2 source ~/micromamba/envs/test2/bin/activate
Conda environments are directories too though... Can you clarify why you're saying this is a problem for uv? Both make bin, lib subdirs in a target directory that have a python with site-packages
`uv venv <.venv or .venv2>` then `source .venv/bin/activate` or `source .venv2/bin/activate` Is this enough for you?
Posted from @napari.fosstodon.org.ap.brid.gy with #napariBluesky.
Thanks for organizing this #TeamTomo challenge, CZII! Let’s train models to pick all the things! 🤖🔬 🧪🧶🧬
CZII’s cryoET particle picking machine learning challenge is up and running. More than 3400 signups on Kaggle, great discussions going on, and leaderboard scores better than we were able to do! Start off at: cryoetdataportal.czscience.com/competition
If folks aren't aware (especially bioimaging folks), we have a cryoET particle picking competition running (with $$$ prizes): cryoetdataportal.czscience.com/competition We're at >3k entrants with >1k submissions right now. 2 months to go, and the leaderboard is quite interesting already!
Hi all, I'm a computer scientist. Currently I'm focused on computational work that connects to bioimaging data. You can find more about me at my personal page kyleharrington.com