Tengfei Liu

[sk @compchemm]

isn't that crazy! 2000 binders with Average_dG(rosetta) Lower than -100. Some uniprot ids; https://t.co/Xlm4avLWyV
https://t.co/7YqiRfbZak
https://t.co/4XgnpClSJr

[nature @Nature]

Nature research paper: Miniaturizing and modifying natural proteins with Raygun

https://t.co/lQ2EcsQfpJ

[nature @Nature]

Nature research paper: An ancient mitochondrial program tunes translation to haem availability

https://t.co/m91pLGTjzv

[phil @1qUXHAB3F2y3sJ8]

@realchendahuang https://t.co/x9E0Hb0rjy互相得5美刀

[ollama @ollama]

DeepSeek-V4-Flash-0731 is Ollama's fastest growing model ever in token usage. We are scaling capacity in US & Europe.

On Ollama, this model runs with high performance (100tps+) and zero data retention. Your data stays yours.

ollama run deepseek-v4-flash:0731-cloud

[Caixia Gao @CaixiaGaoLab]

I'm excited to share our newest publication in @ScienceMagazine . We report the use of AI for the programmable design of synthetic plant immune receptors (SPIRs).
https://t.co/iuiQ6KqkGs

[David R. Liu @davidrliu]

Today in @Nature we report how AI-guided redesign enhances protein evolution. Integrating ProteinMPNN sequence design with autonomous laboratory evolution, we establish a workflow to engineer enzymes with improved properties over those evolved from natural proteins. Redesigned starting points consistently evolve an expanded fitness landscape, reaching new function with higher activity, specificity, and stability than their natural counterparts.
https://t.co/pNlw6z1zvU
1/14

[nature @Nature]

Researchers are rewriting the chemical tags on DNA to tune gene expression

https://t.co/TprECNmJhI

[Yutaka Saito @ytksai]

論文がBioinformatics誌に掲載されました。
既存の相同タンパク質の実験データを、大規模な追加実験なしに別のタンパク質の変異効果予測へ再利用する、新しいデータ拡張法を提案しました。
https://t.co/9iNEW1rBkp

[Ah-Ram Kim @ahramkim1128]

Understanding language model scaling for protein fitness prediction https://t.co/Sm3f9hIR57

[铁柱AGI @cgnot996]

感觉 Google 在憋大招了，先准备一手，太便宜了

万一 Gemini 3.5 Pro 强力估计要涨飞起来

[Ah-Ram Kim @ahramkim1128]

Protein stability is determined by single-site bias rather than pairwise covariance https://t.co/qikjh1VeyP

[Ruihan Guo @RuihanGuo2]

Excited to share our #ICML2026 paper MutAtlas.

We build the first PDB-scale multi-source in-silico protein mutation effect atlas, covering 80k+ protein chains, 15.9M residue positions, and 300M+ single-site mutation evaluations.

MutAtlas brings together energy-based, language-model-based, and inverse-folding-based mutation signals in a unified residue-level preference space, providing a reusable community resource for analyzing and learning from heterogeneous mutation signals.

We further propose disagreement-aware preference distillation, which uses MutAtlas to fine-tune a protein language model without experimental labels and achieves SOTA zero-shot performance on ProteinGym among evaluated sequence-only methods.

ICML 2026 poster: Jul 9 · 10:00–11:45 AM CEST · Hall A #707

[Maximiliano Firtman @firt]

Claude Science opens a web server in your computer and run as a local web app. That's new for AI tools.

[Josh Tycko @JoshTycko]

We hope this protocol is useful for people doing large scale protein function work! https://t.co/IcKt12C5Jg

Led by @abbythurm from @BintuLab

[Andre Watson 🧬 @nanogenomic]

You can ask Claude Science to install LigandAI (pip install ligandai) and reference https://t.co/wPhPK1PyZx to generate nearly instant peptides and come up with target discovery :)

[sk @compchemm]

Protenix-v2 weights available now😍
apples to apples comparison from Lucas Nivon post. Protenix ahead of all; AF3, Boltz, OF3-p2 on a blind test set of protein/ligand. So for Co-Folding use minimum Protenix-v2
https://t.co/nKQvZRzrth

[Asa @app_sail]

https://t.co/CDD9AmB3Pn

[Ben Fry @benf549]

My first-author paper describing our protocol for the zero-shot de novo design of drug-binding proteins is now available as an article in Nature! Here’s what we did and what's new from the preprint posted last year 🧵 (1/10):

[Arc Institute @arcinstitute]

The future of biology is agentic. We're proud to work with NVIDIA on the Evo series of models and are excited to see the NVIDIA BioNeMo Agent Toolkit launch to accelerate programmable biology.