Flow Matching for Optimal Reaction Coordinates of Biomolecular System

2024-08-30

Journal of Chemical Theory and Computation

AI Summary

This study addresses two key challenges in biomolecular kinetics: the difficulty of identifying reactive sites in small molecules and the inaccuracy of reaction coordinate construction. We propose FMRC, a Flow Matching-based Reaction Coordinate learning algorithm. Leveraging principles of reversibility and decomposability, FMRC constructs a conditional probability framework and employs deep generative models to implicitly learn an optimal low-dimensional reaction coordinate for reversible dynamics—marking the first application of flow matching to reaction coordinate learning. Crucially, it encodes dominant eigenfunctions of the transfer operator without explicit operator estimation. Theoretical analysis and rigorous evaluation via Markov State Model (MSM) quality metrics jointly validate its reliability. Across three complex biomolecular systems, FMRC significantly improves MSM accuracy and successfully guides bias potential construction in enhanced sampling. This work establishes a novel paradigm for kinetic modeling and free energy computation.

Technology Category

• Machine Learning: Learning with Manifolds
• Reasoning under Uncertainty: Stochastic Optimization
• Intelligent Robots: State Estimation

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

We present flow matching for reaction coordinates (FMRC), a novel deep learning algorithm designed to identify optimal reaction coordinates (RC) in biomolecular reversible dynamics. FMRC is based on the mathematical principles of lumpability and decomposability, which we reformulate into a conditional probability framework for efficient data-driven optimization using deep generative models. While FMRC does not explicitly learn the well-established transfer operator or its eigenfunctions, it can effectively encode the dynamics of leading eigenfunctions of the system transfer operator into its low-dimensional RC space. We further quantitatively compare its performance with several state-of-the-art algorithms by evaluating the quality of Markov state models (MSM) constructed in their respective RC spaces, demonstrating the superiority of FMRC in three increasingly complex biomolecular systems. In addition, we successfully demonstrated the efficacy of FMRC for bias deposition in the enhanced sampling of a simple model system. Finally, we discuss its potential applications in downstream applications such as enhanced sampling methods and MSM construction.

Problem

• Optimal reaction site
• Biological small molecules
• Flow coordination method

Innovation

• FMRC
• Deep Learning
• Molecular Dynamics