Breaking timescales with generative sampling of conformational transitions.

Gen-COMPAS enables efficient sampling of biomolecular transition pathways within nanoseconds to microseconds, surpassing traditional methods by reconstructing key states without predefined variables.

• Why it matters: Understanding molecular transitions like protein folding and membrane transport is crucial for biological insight but challenging due to their rarity and computational demands, limiting current simulation approaches.
• What they did: The approach combines a denoising diffusion probabilistic model with committor-based filtering, generating plausible intermediates and identifying transition states from end-point structures without prior mechanistic knowledge.
• The result: This method accurately recovers transition states, committors, and free-energy landscapes across diverse systems, offering a practical tool for studying complex biomolecular processes previously difficult to simulate.