Symmetry-Driven Discovery of Dynamical Variables in Molecular Simulations

Jeet Mohapatra, Nima Dehmamy, Csaba Both, Subhro Das, Tommi Jaakkola

TL;DR: We relate effective degrees of freedom to symmetries and use symmetry discovery find important DoF such as dihedral angles in alanine dipeptide

Abstract: We introduce a novel approach for discovering effective degrees of freedom (DOF) in molecular dynamics simulations by mapping the DOF to approximate symmetries of the energy landscape. Unlike most existing methods, we do not require trajectory data but instead rely on knowledge of the forcefield (energy function) around the initial state. We present a scalable symmetry loss function compatible with existing force-field frameworks and a Hessian-based method efficient for smaller systems. Our approach enables systematic exploration of conformational space by connecting structural dynamics to energy landscape symmetries. We apply our method to two systems, Alanine dipeptide and Chignolin, recovering their known important conformations. Our approach can prove useful for efficient exploration in molecular simulations with potential applications in protein folding and drug discovery.

Lay Summary: This research introduces a new way to identify the most important molecular movements in computer simulations of proteins and other molecules. Our approach works by examining the energy landscape around a molecule's starting position and identifying patterns of symmetry - essentially, finding directions where the molecule can move without significantly changing its energy. The key innovation is that your method doesn't need extensive simulation data upfront. It only requires knowledge of the forces acting on the molecule at its initial state. To test our approach, we applied it to two well-studied protein systems: alanine dipeptide and Chignolin. In both cases, our method successfully identified the important molecular conformations that were already known from previous studies. This technique could help researchers explore the conformational landscape without requiring long simulation runs.

Application-Driven Machine Learning: This submission is on Application-Driven Machine Learning.

Primary Area: Applications->Chemistry, Physics, and Earth Sciences

Keywords: AI4Science, Molecular Dynamics, Second Order Methods, Symmetry Discovery

Submission Number: 5055