From snapshots to ensembles: Integrating experimental data and dynamics

Abstract

Protein function arises from the interplay of structure, dynamics, and biomolecular interactions. Despite advances in cryo-EM and AI-based structure prediction, capturing dynamic and energetic features remains a challenge. Biophysical methods like NMR, EPR, HDX-MS, SAXS, and cryo-EM provide valuable but often indirect signals. Connecting these to molecular mechanisms requires integrative approaches that combine experiments with physics-based simulations, revealing both stable structures and transient, functionally important intermediates. This review highlights recent advances in integrative modeling using the maximum entropy principle to build dynamic ensembles from diverse data while addressing uncertainty and bias. These methods help resolve heterogeneity and interpret low-resolution data. We conclude by exploring how integrative modeling, enhanced sampling, and AI-driven tools enable new insights into slow, large-scale conformational changes.