Nick Boyd's Avatar

Nick Boyd

@nboyd.bsky.social

71 Followers  |  148 Following  |  18 Posts  |  Joined: 08.09.2024  |  2.2115

Latest posts by nboyd.bsky.social on Bluesky

This paper is incredible; IMO large-scale, unbiased data is our best chance to get beyond the relatively small set of sequences that AlphaFold loves

17.10.2025 19:24 β€” πŸ‘ 4    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0
Proteinbase The home of protein design data

proteinbase.com from @adaptyv.bio looks amazing - hundreds of open-source binding affinity measurements. It’s hard to overstate the value of high-quality, uniform affinity data for training and evaluating filtering and ranking methods

03.10.2025 23:19 β€” πŸ‘ 2    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Pretty interesting that AFAICT the filtering was done after the fact (so, library 1 had no filtering). This could make it an excellent dataset for training/testing filters/rankers. Too bad it looks like the dataset is not public

29.09.2025 19:40 β€” πŸ‘ 1    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0

Another great post on binder design from @btnaughton.bsky.social !

29.09.2025 01:16 β€” πŸ‘ 1    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

It's a very cool result but IMO there are caveats. Inference is (mostly) slower. There is existing work on faster models (e.g. MiniFold or protenix mini), and also existing work on ensemble prediction. I doubt this works without training on AFDB, which bakes in inductive bias from triangle layers

26.09.2025 18:00 β€” πŸ‘ 10    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0
Preview
GitHub - escalante-bio/mosaic: composite-objective protein design composite-objective protein design. Contribute to escalante-bio/mosaic development by creating an account on GitHub.

could be fun to implement this in github.com/escalante-bi... -- and then try swapping AF2 for Boltz or Protenix, adding an inverse folding term during design, etc

24.09.2025 15:57 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

Great nanobody design paper from @brianhie.bsky.social and @synbiogaolab.bsky.social ! Amazing what combining a few public models can do

24.09.2025 15:54 β€” πŸ‘ 3    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0
Video thumbnail

Today we're dropping the "beta" tag from Adaptyv, launching our new website and announcing our $8M seed round.

When we started Adaptyv a few years ago, our core belief was: AI models for biology are only as good as the lab data they're trained on and the hypotheses they can test in the real world.

18.09.2025 17:40 β€” πŸ‘ 8    πŸ” 3    πŸ’¬ 1    πŸ“Œ 0

Looks interesting (particularly for in silico affinity maturation) & code available on github: github.com/TencentAI4S/...

16.09.2025 21:59 β€” πŸ‘ 9    πŸ” 4    πŸ’¬ 0    πŸ“Œ 0

mpnn actually improved the hit rate and affinities for af2 -- motivation same as af2cycler πŸ˜€. I wouldn't read much into boltz results, there are a ton of hyperparameter choices for boltz we made (e.g. # of diffusion steps, MSAs vs templates, refolding with AF2 multimer) that complicate the story

03.09.2025 19:01 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0

οΏΌFor these designs we used the proven BindCraft filters. `mosaic` is similar to BindCraft but supports many more models (AF2, Boltz 1+2, Protenix, several PLM and inverse folding models, any JAX model). We’re really excited about `mosaic` for more complex design objectives like dual target binders.

02.09.2025 14:40 β€” πŸ‘ 7    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Post image

Recently tested some de-novo minibinders against two targets (thanks Adaptyv!) designed using our open-source design library, `mosaic`; our best method got hit rates of 7/10 and 8/10 and affinities as low as single-digit nanomolar. Wrote up some thoughts here: blog.escalante.bio/minibinder-d...

02.09.2025 14:39 β€” πŸ‘ 38    πŸ” 15    πŸ’¬ 2    πŸ“Œ 1
BindCraft: one-shot design of functional protein binders
YouTube video by Boston Protein Design and Modeling Club BindCraft: one-shot design of functional protein binders

Have no fear if you weren't in the packed audience this week for @martinpacesa.bsky.social's awesome seminar, because you can check out the recording πŸ‘‡
youtu.be/qQihl6If9vU

15.08.2025 20:56 β€” πŸ‘ 24    πŸ” 8    πŸ’¬ 0    πŸ“Œ 0
Machine Learning Researcher in Protein Design (f/m/*) Join a scaleup researching and operating ML-guided lead optimization of proteins. This means developing a combination of protein language models, and multi-property prediction and generation.

✨Cradle is hiring protein+ML researchers!✨

We operate ML for lab-in-the-loop lead optimization across all industries (pharma, synbio, ...), modalities (antibodies, enzymes, ...), properties (binding, activity, ...)

We're a scaleup and already relied upon by 4 of the top 20 big pharma.

Apply here!

18.06.2025 13:55 β€” πŸ‘ 10    πŸ” 6    πŸ’¬ 1    πŸ“Œ 0

For this target it should be, I've run it on an A10 with 24gb

14.06.2025 15:52 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 0    πŸ“Œ 0
Preview
boltz-binder-design/examples/boltz_notebook.py at main Β· escalante-bio/boltz-binder-design multi-objective protein design. Contribute to escalante-bio/boltz-binder-design development by creating an account on GitHub.

Try it out yourself with a demo marimo notebook: github.com/escalante-bi...

Still a work in progress; feedback and PRs welcome :)

p.s. this is a very small target but each one of these designs takes only 14 seconds to generate on an H100 (slightly longer to get discrete sequences)

12.06.2025 20:52 β€” πŸ‘ 1    πŸ” 1    πŸ’¬ 1    πŸ“Œ 0
Video thumbnail

We added Boltz-2 to our protein design package! Under the hood this relies on a JAX translation, which, thanks to @jeremywohlwend.bsky.social and @gcorso.bsky.social ’s clean code, was fairly easy to write. We’ve been getting great results -- and we have yet to explore the most exciting new features

12.06.2025 20:52 β€” πŸ‘ 21    πŸ” 5    πŸ’¬ 1    πŸ“Œ 0
Video thumbnail

Designed my (the?) first minibinder using @gcorso.bsky.social 's new Boltz-2 model. Excited to explore how templates and increased MSA dropout impact binder design. This is work in progress but will be open sourced as usual

09.06.2025 18:01 β€” πŸ‘ 11    πŸ” 2    πŸ’¬ 0    πŸ“Œ 1
boltz-binder-design/examples/energy.py at main Β· escalante-bio/boltz-binder-design multi-objective protein design. Contribute to escalante-bio/boltz-binder-design development by creating an account on GitHub.

Interesting! Seems very similar to what @sokrypton.org does with his contact losses; probably encourages all mass in one PAE bin? I added it to `boltz-binder-design` for boltz-1 and AF2: github.com/escalante-bi...

31.05.2025 19:45 β€” πŸ‘ 4    πŸ” 1    πŸ’¬ 0    πŸ“Œ 0
add esmj (and joltz and boltz-binder-design :) ) to awesome-list.md by nboyd Β· Pull Request #1020 Β· patrick-kidger/equinox Hi Patrick, Adding ESM-C as requested! I also shamelessly slipped in joltz and boltz-binder-design -- can remove these if you like Nick

Sure, would love to! Just submitted a PR

09.05.2025 16:44 β€” πŸ‘ 0    πŸ” 0    πŸ’¬ 1    πŸ“Œ 0
Preview
GitHub - escalante-bio/esmj: jax translation of esmc jax translation of esmc. Contribute to escalante-bio/esmj development by creating an account on GitHub.

Want to try ESMC but don't want to switch from #JAX to pytorch?

Here’s a super barebones translation to @patrickkidger.bsky.social's excellent equinox: github.com/escalante-bi...

08.05.2025 19:53 β€” πŸ‘ 5    πŸ” 1    πŸ’¬ 1    πŸ“Œ 2
Preview
GitHub - nboyd/joltz: JAX translation of boltz JAX translation of boltz. Contribute to nboyd/joltz development by creating an account on GitHub.

No chunking, but this would be very easy to shard… github.com/nboyd/joltz

14.12.2024 16:36 β€” πŸ‘ 5    πŸ” 0    πŸ’¬ 2    πŸ“Œ 0

@nboyd is following 19 prominent accounts