Reposted by @jurgjn.bsky.socialAdam Kucharski @adamjkucharski.bsky.social · 20/07/2026“I got Claude Fable to do this analysis and write this paper. I’m posting it here, but I haven’t checked it yet.” WELL CHECK IT THEN. WHY ARE YOU POSTING THINGS YOU CAN’T BE BOTHERED TO READ. 47812
Reposted by @jurgjn.bsky.socialAmir Bitran @amirbitran.bsky.social · 05/07/2026I am delighted to share that my newest manuscript and the first of my postdoc in the Marqusee and Bustamante labs, “A detailed molecular picture of protein folding during active translation”, is now on the BioRXiv! (1/n) www.biorxiv.org/content/10.6...biorxiv.orgA detailed molecular picture of protein folding during active translationAll proteins can begin to fold on the ribosome, and many proteins critically rely on co-translational folding to attain their native conformation. The molecular details underlying this crucial process... 24211
Reposted by @jurgjn.bsky.socialHarmit Singh Malik @harmitmalik.bsky.social · 24/06/2026I know it seems like a heretical idea, but one way to figure out if a preprint is really interesting (to you) is... to read the preprint. Then, in the spirit of sharing, send a note to authors: a simple kudos or a more detailed question. I do this a dozen times a year. It's awesome. Try it. 1229264
Reposted by @jurgjn.bsky.socialMarc Robinson-Rechavi @marcrr.bsky.social · 31/03/2026This! AlphaFold is PDB + UniProt + Deep learning. No data sharing, no curation, no data standards, no AlphaFold. 23517
jurgjn.bsky.social @jurgjn.bsky.social · 31/03/2026We've updated our pooled co-folding map of yeast protein-protein interactions. We now include 7,631,089 heterodimers, increasing coverage from 28.9% to 63.4% of all possible pairs. The web app with all the structures has grown to 15GB, but should still run locally! github.com/jurgjn/poole...github.comGitHub - jurgjn/pooled-ppi-yeast: Work-in-progress map of yeast protein-protein interactions using pooled co-foldingWork-in-progress map of yeast protein-protein interactions using pooled co-folding - jurgjn/pooled-ppi-yeast 052
Reposted by @jurgjn.bsky.socialMohammed AlQuraishi @moalquraishi.bsky.social · 13/03/2026New OpenFold3 preview out! (OF3p2) It closes the gap to AlphaFold3 for most modalities. Most critically, we're releasing everything, including training sets & configs, making OF3p2 the only current AF3-based model that is functionally trainable & reproducible from scratch🧵1/9 124490
Reposted by @jurgjn.bsky.socialPedro Beltrao @pedrobeltrao.bsky.social · 04/03/2026We have started a project trying to predic the interactions/structures of all yeast protein pairs using an AlphaFold pooling approach. We are making the current dataset open and we welcome collaborations. www.evocellnet.com/2026/03/mapp...evocellnet.comMapping the yeast atructural interactome with AlphaFold3: an open call for collaborationWe are excited to announce the early-stage release of our S. cerevisiae structural interactome mapping project. Using AlphaFold3 (AF3), w... 69853
Reposted by @jurgjn.bsky.socialAndre Cornman @ancornman1.bsky.social · 03/03/2026Predicting protein-protein interactions (PPIs) at proteome scale can take months with co-folding models due to the massive all-vs-all comparisons required. We are excited to announce FlashPPI, a contrastive learning framework that predicts proteome wide physical interfaces in minutes. 1/🧵 16727
Reposted by @jurgjn.bsky.socialLeonardo Carella @leonardocarella.bsky.social · 24/02/2026Slava Ukraini. 0223
Reposted by @jurgjn.bsky.socialChris Ingraham @cingraham.bsky.social · 18/02/2026Elon posted this yesterday. Just for fun I did it with my before/after cancer PET scans that I had already posted online. In the "before" scan Grok missed the massive tumor lighting up my liver, and in both scans it flagged a non-existent "area of concern" in my chest. Other than that, works great! 833042609
Reposted by @jurgjn.bsky.socialCarl T. Bergstrom @carlbergstrom.com · 30/10/2025If you want to implement an AI tutor, you need to study it’s performance in naturalistic settings with the kinds of context windows that student learners will provide. You can’t begin with a well formulated question about a single topic. Students don’t know how to formulate questions well at first. 412612
Reposted by @jurgjn.bsky.socialMichael Hoffman @michaelhoffman.bsky.social · 22/10/2025@drkatemarvel.bsky.social Are there any good current equivalents of the red stack from David MacKay 2009? It lives rent-free in my head but would like to know if it still reflects current knowledge. www.withouthotair.com/c18/page_103... 141
Reposted by @jurgjn.bsky.socialDrug Monkey @drugmonkey.bsky.social · 25/09/2025This may be the most important paper ever published about NIH funded research. www.science.org/doi/10.1126/...science.orgWhat if NIH had been 40% smaller?Replaying history with less NIH funding shows widespread impacts on drug-linked research 5414250
Reposted by @jurgjn.bsky.socialJeffrey Barrett @jeffbarrett.eu · 24/09/2025The science is, in fact, clear. Good studies [1] consistently show a weak association between tylenol & autism. Those studies are very careful to say they can't establish whether this is causal. A huge Swedish study used a clever sibling design to address this, and showed zero causal effect [2]. 22610
Reposted by @jurgjn.bsky.socialPedro Beltrao @pedrobeltrao.bsky.social · 22/07/2025Unexpectedly, @jurgjn.bsky.social found that running Alphafold3 predictions for protein interactions can yield ipTM scores that are more predictive of true interactions when run in pools of proteins instead of pairwise predictions. Presumably, this reflects some sort of "competition effect". 38731
Reposted by @jurgjn.bsky.socialDavid Pfau @davidpfau.com · 28/05/2025The war on science in the US is already having an effect on private sector research like AlphaFold. Bears repeating but the private sector builds on top of things created by academic research for the public good. This hurts everyone. 2236103
Reposted by @jurgjn.bsky.socialDiego del Alamo @delalamo.xyz · 23/04/2025Some crystal structures show evidence of multiple stable conformations, such as loops or side chains with two distinct regions of density. These can be correctly modeled by MD, but not modern protein ensemble prediction tools like AlphaFlow or DiG (BioEmu not tested) www.biorxiv.org/content/10.1... 04412
Reposted by @jurgjn.bsky.socialArjun Raj @arjunraj.bsky.social · 20/03/2025I miss when the biggest controversy about mRNA was whether it correlated with protein or not. 1031639
Reposted by @jurgjn.bsky.socialCarl T. Bergstrom @carlbergstrom.com · 23/02/2025LLMs can't take responsibility for their mistakes. When a human journalist puts their name on AI-written text, they take on that responsibility. Increasingly I see inaccurate and badly written news stories authored by AI, many of which have actual humans listed as authors or editors. 625635
Reposted by @jurgjn.bsky.socialTimothée Poisot @ctrlalttim.com · 05/12/2024The best first sentence of a grant application I've read was (paraphrasing), "Tool X is widely used to do task Y; we will make it accessible to people living with condition Z (13% of the population) so that it is more equitable and more widely used". Let's unpack why, because there's a lesson. 🧵 2522239
Reposted by @jurgjn.bsky.socialjurgjn.bsky.social @jurgjn.bsky.social · 18/11/2024We’ve now explored this on all GPU models available on our cluster and some non-A100/H100 do not have the issue; seems like something nuanced that can hopefully be fixed: github.com/google-deepm...github.comModel does not run correctly on non A100/H100 GPUs · Issue #59 · google-deepmind/alphafold3A note from us at Google DeepMind: We have now tested accuracy on V100 and there are serious issues with the output (looks like random noise). Users have reported similar issues with RTX 2060S and ... 035
Reposted by @jurgjn.bsky.socialPedro Beltrao @pedrobeltrao.bsky.social · 14/11/2024Jurgen in our group run AF3 locally after fiddling. It worked well on a A100 and produced some bad models on a lower end GPU. We don't have access to many A100s so unless we get it working on lower-end GPUs we won't be able to use this much. 2172
Reposted by @jurgjn.bsky.socialCarl T. Bergstrom @carlbergstrom.com · 20/10/20241. I'm conflicted about moving old tweets to Bluesky. I like the idea in general of using @blueark.app to preserve that content. I have a few threads posted during COVID—for example, the original analysis of the IHME model right after it came out—that seem a valuable part of the scientific record.blueark.appBlueArkThe new blue land is waiting for you. Bring your stories with BlueArk. 1111513
