Claude autonomously executes science workflows, beats standard protein design hit rates
Dr_Singularity · x · 2026-08-19
Anthropic's Claude is performing real experimental science workflows almost autonomously. In a protein design campaign, it generated binders for 14 of 15 targets, achieving hit rates of 23–35% compared to the typical 10–15%, with some designs matching or beating the best previously reported binding affinities. It handled the entire computational pipeline itself, including selecting binding sites.
Related event: Anthropic Says Claude Designed Protein Binders for 14 of 15 Targets(45 posts)→
More from Research
- Five pharma giants federally train OpenFold3 on 20,000 private structures, +11 pts — jiqizhixin · 2026-10-16
- SWAN: a semantic-level text watermark that survives paraphrasing — burkov · 2026-10-16
- Interp researchers push back: 'networks are already clean' claim rests on dropout artifact — aryaman2020 · 2026-10-16
- LayerTrim: training-free pruning keeps reasoning LLMs intact on AIME — SeoulNatlUniv · 2026-10-16
- Simular presents neuro-symbolic computer use agents at BayLearn — xwang_lk · 2026-10-16
- Google's Hartmut Neven disputes chip test, outlines plan to link a brain to a quantum computer — Chris_Armstrong · 2026-10-16