I'm really proud of this work by Jae Kennemur and Yueming Long. #Enzymes rock, with some help from #directed evolution to do chemistry not known in nature (and really hard for humans, too). @caltechcce.bsky.social pubs.acs.org/doi/10.1021/...

The posting is unfortunately not super clear on this, but please apply if you want to do BioML research, especially around machine learning for molecular biology and bioengineering! I get sad if there are no applicants with this profile in the pool!

Very excited about our latest all-atom generative model proteina, check out the project page (research.nvidia.com/labs/genair/...) and stay tuned for the code release soon!

Really nice metastudy of citizen science contributing to ecology & biodiversity research @inaturalist.bsky.social @aibsbiology.bsky.social 👀🔎🍄🧪🌎🌐 iNaturalist accelerates biodiversity research academic.oup.com/bioscience/a...

The MLSB workshop will be in San Diego, CA (co-located with NeurIPS) this year for its 6th edition in December 🧬🔬 Stay tuned @workshopmlsb.bsky.social as we share details about the stellar lineup of speakers, the official call for papers, and other announcements!🌟

There's a lot more in the preprint! www.biorxiv.org/content/10.1...

In 1965, Margaret Dayhoff published the Atlas of Protein Sequence and Structure, which collated the 65 proteins whose amino acid sequences were then known. Inspired by that Atlas, today we are releasing the Dayhoff Atlas of protein sequence data and protein language models.

“Writing compels us to think — not in the chaotic, non-linear way our minds typically wander, but in a structured, intentional manner. By writing it down, we can sort years of research, data and analysis into an actual story, thereby identifying our main message and the influence of our work.”

🧪 The EPA is eliminating its scientific research arm and firing hundreds of scientists. This move, enabled by a Supreme Court ruling, dismantles the office providing independent research for environmental policies. #Climate

We're sponsoring the use of ESM3 and EMSC to help researchers engineer improved PETase enzymes in the @AlignBio 2025 Protein Engineering Tournament. Get started using ESMC to predict protein function and ESM3 to generate new enzymes here: github.com/evolutionary...

1/4 🚀 Announcing the 2025 Protein Engineering Tournament. This year’s challenge: design PETase enzymes, which degrade the type of plastic in bottles. Can AI-guided protein design help solve the climate crisis? Let’s find out! ⬇️ #AIforBiology #ClimateTech #ProteinEngineering #OpenScience

Unfortunately, the MLSB Workshop @ NeurIPS (@workshopmlsb.bsky.social) was rejected this year. Feedback from the deciding committee indicates it was a coin flip decision, with 283 proposals & a number related to “computational biology” More on the future of MLSB soon…

Three BioML starter packs now! Pack 1: go.bsky.app/2VWBcCd Pack 2: go.bsky.app/Bw84Hmc Pack 3: go.bsky.app/NAKYUok DM if you want to be included (or nominate people who should be!)

Pls tell me im living under a rock and we have citation managers compatible with iPads where your writing & highlighting can be automatically transcribed into notes and key words attached to a pdf (vs me manually transcribing over from edits I make in Adobe??) In the age of AI, we have this, right??

With contributions from fantastic colleagues @martinsteinegger.bsky.social , @mikeinouye.bsky.social, @jlistgarten.bsky.social , @ideasbyjin.bsky.social, @michael-heinzinger.bsky.social, and many more, the first CSHL volume on ML for Protein Science and Engineering is out: lnkd.in/dQdgGPpp

Physics-based design of efficient Kemp eliminases @lynnkamerlin.bsky.social www.nature.com/articles/s41...

All all-atom diffusion model of protein sequences. www.biorxiv.org/content/10.1...

End-to-end differentiable homology search for protein fitness prediction. @yaringal.bsky.social @deboramarks.bsky.social @pascalnotin.bsky.social arxiv.org/abs/2506.089...

🚨 New paper 🚨 RNA modeling just got its own Gym! 🏋️ Introducing RNAGym, large-scale benchmarks for RNA fitness and structure prediction. 🧵 1/9

Which RNAs are functional? Been a long time goal of @deboramarks.bsky.social to start an RNA sequence —> function and structure mapping. Huge thx to team for pushing on this when - led by Rohit Arora and Murphy Angelo for seeing this thru and Pascal Notin for pulling together!!!

RNA, RNA, RNA -> RNAGym! read all about it! biorxiv.org/content/10.110… Been a long time goal of @deboramarks.bsky.social We tried a million years ago, sciencedirect.com/science/articl��� Now total refresh led by Rohit Arora and Murphy Angelo and Pascal Notin for pulling together!!

Happy to release our breakthrough AI-model: RareFold, which predicts and designs proteins with noncanonical AAs. With EvoBindRare, we designed linear & cyclic peptide binders with high affinity & novel binding modes, wet lab validated. 📄 biorxiv.org/content/10.1... 💻 github.com/patrickbryan...

🎉 Congrats to Nate for his awesome preprint! We used deep learning to design phages with complex infectivity and specificity profiles. Big shifts in host targeting come from just a few mutations! Training on multifunctional data enables precise control over protein properties 🧬 tinyurl.com/yc4wtn8h

Super excited to be presenting Dyna-1 next week with @hkws.bsky.social !!

Happening in 15 mins! Message if you need a link :)

Indeed, 1 A.A. mutation from each of these hotspots dramatically extended the pulsatory response (>18 hrs vs 10s of min), suggesting evolutionary tuning of dynamics.

Remarkably, sequence analysis found these *exact same hotspots* had co-evolved in BcLOV4 homologs, implying function.

Structural studies (limited proteolysis, HDX-MS by the Gardner lab) identified temperature-sensitive hot-spots in DUF.

What about the temperature-sensitive inhibitory Node C? We mapped this to the mysterious “Domain of Unknown Function” (DUF). We still call it DUF even though function is now known :P.

We then mapped the network nodes: The light-sensing LOV domain (A) regulates a charged clustering region (B), which drives PM binding. Aside: this inspired an entire toolset for single-protein translocation to various compartments. Preprint soon!

The inter-domain IFFL idea started as a hunch, but models made several non-intuitive predictions that we then confirmed experimentally through mutations that altered the energetics of LOV domain interactions.

Answer: smaller-scale networks of interactions between the protein’s domains. Light response (A) quickly activates a region for plasma membrane (PM) binding (B), but also enables a temp.-sensitive domain (C) to inhibit the PM-binding region more slowly.

Step-to-pulse response – adaptation – is common in all kinds of networks, often through an incoherent feed-forward loop (IFFL). Node A quickly activates Node B but also —more slowly— Node C, which inhibits Node B. Result: a pulse of B. But how can a single protein do it w/o a larger network?

Why does BcLOV4 turn itself off despite constant light stimulation, and why does this depend on temperature? The answer could make better optogenetic and thermogenetic tools, and itself was a fascinating puzzle.

How can an individual protein show adaptation? The light- and temp-sensitive BcLOV4 gives a pulse of translocation during a step input of light. Answer in our preprint: *intra-molecular* feed-forward regulation. bit.ly/3Rg9veD A @bugajlab.bsky.social collab with @nmrkaygee.bsky.social 🧵

Search among all posts in this feed 👉 mlsb.papers.blue

Structure models generate proteins with higher structure confidence and more plausible energy distributions, but limited diversity and sequence biases. Sequence models generate diverse and novel proteins with lower structural confidence. Neither are good at generating enzymes.

RFDiffusion2: a diffusion model that scaffolds enzyme active sites at the atomic level enables the generation of functional de novo enzymes www.biorxiv.org/content/10.1...

Congratulations to Elliott Weix on being named a 2025 Goldwater Scholar! Weix, a biochemistry and math major, works in the @cellraiser.bsky.social Lab. biochem.wisc.edu/2025/04/10/b...

Supervised training using data generated by multiplexed assays of variant effects is potentially very powerful, but is made difficult by assay- and protein-specific effects Here @tkschulze.bsky.social devised a strategy to take this into account while training models www.biorxiv.org/content/10.1...

Our work on better ways of aggregating PLM embeddings is now out in Bioinformatics Advances.

I haven’t hosted in quite some time!! But super excited to be back today and host Anthony as he goes over his new paper out in Nature! Note the 11AM EST time change (so he isn’t up too late on the other side of the globe😉) hope to see people there!

Now just a few days left to submit your abstract for a chance to join this fantastic protein design meeting 📅 📜 🧬 💻

Gene synthesis is often the most expensive part of protein engineering with generative models. Happy to have played a small part in this work, where Chase developed a method for precision library construction at scale, with per-gene costs as low as $1.50. @philromero.bsky.social

We collaborated with @hf.co 🤗 to host all of our models and data online! Big thank you to @danielvanstrien.bsky.social @wauplin.hf.co @lhoestq.hf.co for helping out (and the GPU)! Dyna-1 and RelaxDB on HF: 🚀 huggingface.co/spaces/gelne... 🌟 huggingface.co/datasets/gel...

🎉Congrats to Chase on her new preprint! She developed OMEGA--a simple method for assembling custom gene panels for as little as $1.50 per gene. Big step forward protein engineering and design!🧬 www.biorxiv.org/content/10.1...

Scalable and cost-efficient custom gene library assembly from oligopools https://www.biorxiv.org/content/10.1101/2025.03.22.644747v1

From a colleague in my PhD lab! Chase presents her method OMEGA, a simple, scalable method to assemble 100s-1000s of custom genes from oligo pools using standard lab tools! #synbio #proteinengineering #OMEGA

Here's my interview with @stevenstrogatz.com on @quantamagazine.bsky.social's Joy of Why podcast. This was *so much fun*. I was nervous going into it (I really look up to Steve!), but I had a blast and I think this is the best interview I have ever done. Thanks for having me on, folks!

Next Tues (3/18) at 4PM ET we will have Alan Amin present "Bayesian Optimization of Antibodies Informed by a Generative Model of Evolving Sequences" Paper: arxiv.org/abs/2412.07763 Sign up on our website for zoom links!