Gen-COMPAS: generative sampling breaks molecular timescales in Nature paper
LeoTZ03 · x · 2026-10-05
ML4PE will host Chenyu Tang presenting Gen-COMPAS, a generative committor-guided path-sampling framework published in Nature (vol. 657).
- Molecular transitions like protein folding, allostery and membrane transport are too rare for standard MD, while enhanced sampling is costly and relies on arbitrary collective variables
- Gen-COMPAS couples a denoising diffusion model that proposes structurally plausible intermediates with committor-based filtering to identify transition states, without predefined collective variables
- Talk: Oct 6, 10–11 AM ET
More from Research
- OneCanvas turns a single image into 3D spatial reasoning, hitting new SOTA — Roger_M_Taylor · 2026-10-06
- Hugging Face researcher: encrypted reasoning is the worst thing to happen to LLM evals — _lewtun · 2026-10-06
- WM-VLM: world model generates visual intermediate states for spatial reasoning — ZhitingHu · 2026-10-06
- ProAR turns autoregressive video models into goal-oriented prospective reasoners — muhao_chen · 2026-10-06
- 90+ Claude Opus 5.5 agents find two room-temperature magnetic semiconductor candidates in 3 days — Dr_Singularity · 2026-10-06
- Redditor Post-Trains Yandex AliceAI-80B-A3B From Scratch on 3 Local V100s, Update #5 — jjusko20 · 2026-10-06