Breaking timescales with generative sampling of conformational transitions

TL;DR - A generative, committor-guided path-sampling framework reconstructs rare biomolecular conformational transitions without predefined collective variables or brute-force simulation. It enables analysis of otherwise inaccessible transition pathways, thermodynamics, and kinetics at an acceptable computational cost.

• Uses generative sampling to overcome the long timescales associated with rare conformational events.
• Guides path sampling with the committor rather than relying on predefined collective variables.
• Recovers both transition mechanisms and their underlying thermodynamic and kinetic properties.
• Reduces the need for computationally prohibitive brute-force sampling.