Energy-Based Flow Matching for Generating 3D Molecular Structure

2025-08-26

AI Summary

This work addresses the fundamental problem of 3D molecular conformation generation in computational biology. We propose an energy-guided flow matching framework that models molecular conformations as distributions on an energy-driven manifold, learning a differentiable, iterative mapping from random initial configurations to target conformations via deep neural networks. Crucially, physics-inspired energy functions are explicitly incorporated during both training and inference to ensure idempotency and numerical stability; additionally, an AlphaFold-inspired structural refinement strategy is integrated to enhance geometric validity. Compared to state-of-the-art flow matching and diffusion-based methods, our approach achieves new SOTA performance on protein–ligand docking and protein backbone conformation generation tasks, while maintaining comparable computational cost.

Technology Category

• Machine Learning: Learning with Manifolds
• Search and Optimization: Sampling/Simulation-based Search
• Computer Vision: 3D Computer Vision

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments

Abstract

Molecular structure generation is a fundamental problem that involves determining the 3D positions of molecules' constituents. It has crucial biological applications, such as molecular docking, protein folding, and molecular design. Recent advances in generative modeling, such as diffusion models and flow matching, have made great progress on these tasks by modeling molecular conformations as a distribution. In this work, we focus on flow matching and adopt an energy-based perspective to improve training and inference of structure generation models. Our view results in a mapping function, represented by a deep network, that is directly learned to extit{iteratively} map random configurations, i.e. samples from the source distribution, to target structures, i.e. points in the data manifold. This yields a conceptually simple and empirically effective flow matching setup that is theoretically justified and has interesting connections to fundamental properties such as idempotency and stability, as well as the empirically useful techniques such as structure refinement in AlphaFold. Experiments on protein docking as well as protein backbone generation consistently demonstrate the method's effectiveness, where it outperforms recent baselines of task-associated flow matching and diffusion models, using a similar computational budget.

Problem

• Generating 3D molecular structures from random configurations
• Improving training and inference of flow matching models
• Addressing molecular docking and protein backbone generation

Innovation

• Energy-based flow matching for 3D molecules
• Iterative deep network mapping random configurations
• Theoretically justified with idempotency and stability