Researcher shows Fable predicting protein contact maps straight from sequence, top 512 pairs ranked
sokrypton · x · 2026-11-18
- sokrypton demonstrates an unconventional approach: prompting the protein language model Fable to predict, from a 286-residue sequence alone, residue pairs with C-beta atoms within 8Å and sequence separation ≥6, returning the top 512 pairs ranked by confidence as a contact map image.
- The attached contact map shows coherent diagonal contact clusters, suggesting plausible structure capture without a full structure-prediction pipeline.
- In follow-up discussion, he notes getting per-site logits would require multiple runs per site to build a distribution (including responses to every mutation) — too expensive, and his weekly token budget is exhausted; he muses that an alanine scan might approximate the experiment.
More from Research
- Asking Gemini 10,701 questions: Google Search rewrites ~45% of brand picks — dejanseo · 2026-11-18
- Ai2 brings open OlmoEarth models to global field-boundary mapping effort — allen_ai · 2026-11-18
- Rubin Observatory invites amateur astronomers to apply for LSST research data access — mjuric · 2026-11-18
- Boltz's long-term goal: hypothesis to human-ready molecule without leaving the computer — GabriCorso · 2026-11-18
- Boltz Lab now a daily tool for pharma scientists; partners include Pfizer, GSK, Takeda, BMS — GabriCorso · 2026-11-18
- Boltz raises $110M Series A for AI biomolecular design, partners with 6 top-20 pharma giants — GabriCorso · 2026-11-18