ProteoCraft protein-design framework hits 82.9% wet-lab hit rate, best SPR 0.879 nM
jiqizhixin · x · 2026-11-21
A Stanford-led team with HKUST Guangzhou reports ProteoCraft, a protein-design framework achieving 29 binding hits among 35 expressed PD-L1 miniprotein candidates (82.9% hit rate), with best SPR fit of 0.879 nM.
The method treats candidate sequences as search states:
- Approximate gradients from structure predictors propose promising mutations, combining ESMC sequence priors, pairwise geometry constraints, and coordinate-sensitivity diffusion signals;
- Full structure prediction separately verifies each discrete sequence, with rewards combining pTM, ipTM, binder pLDDT and compactness penalties;
- Tree search runs in two stages: multi-site basin exploration, then single-site PUCT refinement of promising candidates.
The framework supports de novo binder design, TCR-facing binder design, and localized CDR/TCR antibody redesign.
More from Research
- Smaller experiments win: the Kaggle lesson for doing more AI research with less compute — strickvl · 2026-11-21
- Ramez Naam extrapolation: AI a decade away from completing 98% of hour-long human tasks — GaryMarcus · 2026-11-21
- Colgate paper: LLM roleplay hits 90% of human test-retest reliability in purchase-intent research — 2C_ornot2C · 2026-11-21
- Riemann Hypothesis proof gets full formal verification, experts call it 'very human math' — Kyrannio · 2026-11-21
- Claims surface: Navier-Stokes and Riemann proofs via LEAN, by OpenAI and Claude — BasketVolleyball11 · 2026-11-21
- Hassabis predicted 2 years for AI to crack a Millennium Prize problem; 6 months later, both Riemann and Navier–Stokes have proposed solutions — Hesamation · 2026-11-21