'Epitope excision' crops protein targets to slash compute in de novo binder design
anshulkundaje · x · 2026-11-01
A new paper introduces epitope excision: cropping targets to a sequence window around binder contact residues plus a spatial halo, then running RFdiffusion and AlphaFold3 on the crop instead of the full target. Across a 15-target, 3,666-design benchmark it largely preserves quality while cutting tokens and wall-clock time; structural biologists quip the trick is a rediscovery.
More from Research
- TifinaghBench launches: top AI model scores just 29.9/100 on Amazigh language understanding — gekobraa · 2026-11-01
- CoBit paper: one continuous bitstream diffusion recipe spans language, image, speech and protein generation — LucaAmb · 2026-11-01
- awesome-ai-constitutions repo collects AI constitutions & model specs with generalization research — dhadfieldmenell · 2026-11-01
- Karpathy's nanochat hits 59K stars: a full LLM training lab in one repo — techNmak · 2026-11-01
- Same 2.4 eV band gap, different routes: SHAP explains per-prediction feature contributions — bravo_abad · 2026-11-01
- Hands-on tutorial trains a decision model on Qwen3-0.6B from scratch — dl_weekly · 2026-11-01