Super cool!! Do you have any insight on the compute boost in comparison to FastRelax (I believe it’ll be massive) and also how well the scores co-relate?

Exciting to see our protein binder design pipeline BindCraft published in its final form in @Nature ! This has been an amazing collaborative effort with Lennart, Christian, @sokrypton.org, Bruno and many other amazing lab members and collaborators.

www.nature.com/articles/s41...

🚀 Excited to release BoltzDesign1!

✨ Now with LogMD-based trajectory visualization.
🔗 Demo: rcsb.ai/ff9c2b1ee8
Feedback & collabs welcome! 🙌

🔗: GitHub: github.com/yehlincho/Bo...
🔗: Colab: colab.research.google.com/github/yehli...
@sokrypton.org @martinpacesa.bsky.social

Love to see entropy getting the attention it deserves ha! Great read to start Monday morning.

Cool work from Profluent on scaling up protein language models ft. ProGen-3!

If anyone has any tips for fellowships or grants or giftfunds our group can apply for (in the US), please send our way.

So far, we've not had much luck in raising funds... 💰 In this relatively popular area, so I feel like I'm doing something wrong 😅

Excited to share our preprint “BoltzDesign1: Inverting All-Atom Structure Prediction Model for Generalized Biomolecular Binder Design” — a collaboration with
@martinpacesa.bsky.social, @Zhidian Zhang, @Bruno E. Correia, and @sokrypton.org

🧬 Code will be released in a couple weeks

BoltzDesign1: Inverting All-Atom Structure Prediction Model for Generalized Biomolecular Binder Design Deep learning in structure prediction has revolutionized protein research, enabling large-scale screening, novel hypothesis generation, and accelerated experimental design across biological domains. R...

Woke up to have #BoltzDesign1 on the top of my reading list. Was excited since I heard about it from Sergey in Keystone!
Work by @yehlincho.bsky.social et al.

www.biorxiv.org/content/10.1...

Screencap from the homepage showing various structural similarity metrics between a pose and a reference structure

A single python package for calculation of quality metrics like DockQ, LDDT, RMSD, & others

peppr.vant.ai

So much to process and be excited about! Massive respect to the organizers and speakers for an incredible Keystone experience! 🚀 #KSMLStructure2025

Also learnt a lot from @elanasimon.bsky.social talk on interpretable features in LLMs, and @possuhuanglab.bsky.social and Bruno Correia’s talks on de novo protein design and the long term vision of ProteinML!

Finally, a wonderful closing keynote by Charlotte Deane highlighted that “AI Is Not Magic”
Charlotte reminded us that structures tell stories—and that understanding beats leaderboard chasing.
She also demoed MolSnapper, a tool to tailor plausible molecules into binding sites.

In two separate talks, @moalquraishi.bsky.social and @sokrypton.org unpacked what AlphaFold has learnt and how it learns, as well as, how to hack it for protein design tasks! (s/o to BindCraft by @martinpacesa.bsky.social et al.)

“ML Methods for Protein–Peptide Interactions” by Amy Keating!
A killer comparison of structure predictors (AF3, Boltz, Chai-1) on 631 protein–peptide complexes! We also got a preview of COORdinator2, a new coordinate-based Potts model that maps sequence landscapes on fixed backbones.

“Decoding Binding Entropy Through Structural Biology” by @stephanieaw.bsky.social
Brilliant exploration of entropy’s role in binding thermodynamics.
Key insight: While binding sites lose entropy, distal sites compensate by becoming more flexible. Very curious how this plays out in Ab–Ag binding!

“Have Protein Language Models Learned Dynamics?” by @hkws.bsky.social As someone whose PhD focused on conformational dynamics, I loved this talk. Hannah showcased the Dyna-1 model that uses NMR data + PLMs to extract the ground truth of protein dynamics!

Just wrapped the @keystonesymposia.bsky.social conference last night – still buzzing from all the spicy talks and incredible science! 🌶️✨
Here’s a quick thread of my personal favorites from the cutting edge of protein ML, dynamics, and design:

Just arrived in Keystone,CO for the @keystonesymposia.bsky.social conference on ML Applied to Macromolecular Structure and Function! Super excited to nerd out on all things proteins + ML.
If you’re around, hit me up — would love to catch up with familiar faces and meet new ones! #KSMLStructure25

+1 to the first part of the statement; maybe with the caveat that if it is templated off a single sequence, the designed sequence has a novel structural fold. Also, what about trastuzumab with redesigned CDRs?