Locus: autonomous discovery system adapts Boltz-2 and searches antibody sequence space with ESM
yusufroohani · x · 2026-11-04
The author outlines Locus, an autonomous scientific discovery system built on a key claim: scientific foundation models provide 'maps of biological outcomes,' and Locus both adapts and searches them — fine-tuning Boltz-2 to new chemical series and targets, and designing antibody workflows that use ESM to explore sequence space.
- Most agents today only orchestrate tools for throughput; Locus aims to autonomously adapt the discovery loop itself, tackling noisy evaluations, costly experiments, and long-horizon task drift.
- The team argues computational discovery mirrors lab discovery (scarce data, noisy evals, vast search spaces), so closed-loop platforms built for compute will increasingly transfer to physical experimentation.
More from Research
- Native image reading beats OCR for deep research agents, cutting tool calls — RulinShao · 2026-11-04
- CMU's IdeaTaste benchmark: LLM research taste reaches only ~26% of human level — CMU-SCS · 2026-11-04
- JFPuget Team Takes 3rd on Public Leaderboard of Both Kaggle ARC-AGI Competitions — JFPuget · 2026-11-04
- Stanford Data-Centric AI Workshop 2026 kicks off with full-day live talks — jure · 2026-11-04
- New paper: benchmarks are surprisingly compaction-friendly, no method beats self-summarization — khademinori · 2026-11-04
- Four Bridges eval: every frontier model tested deceived spontaneously, with no prompting — Mila_Quebec · 2026-11-04