Learning Structure, Energy, and Dynamics: A Survey of Artificial Intelligence for Protein Dynamics

TL;DR

This survey reviews recent AI methods for understanding and simulating protein dynamics, addressing challenges like data scarcity and computational cost through various machine learning approaches.

Contribution

It provides a comprehensive overview of AI techniques for protein dynamics, highlighting recent advances, datasets, and open challenges in the field.

Findings

Summarizes methods for conformation and trajectory generation.

Discusses physics-aware machine learning potentials.

Identifies key challenges like scalability and thermodynamic consistency.

Abstract

Protein dynamics underlie many biological functions, yet remain difficult to characterize due to the high computational cost of molecular dynamics simulations and the scarcity of dynamic structural data. This survey reviews recent advances in artificial intelligence for protein dynamics from three perspectives: learning from structural ensembles and trajectories, learning from physical energy signals, and learning to accelerate molecular simulations. We summarize representative methods for conformation ensemble generation, trajectory generation, Boltzmann generators, physics-aware adaptation, machine learning potentials, coarse-grained modeling, and collective variable discovery. We further discuss available datasets and key open challenges, such as scalability, thermodynamic consistency, kinetic fidelity, and integration with experimental constraints.