Energy-Guided Flow Matching Enables Few-Step Conformer Generation and Ground-State Identification

2025-12-27

AI Summary

Addressing the challenge of balancing conformational diversity and energy reliability in molecular conformation generation—and the lack of ensemble-based modeling capability for ground-state identification—this paper proposes Energy-guided Flow Matching (EnFlow). EnFlow couples an explicit, differentiable energy function with a non-Gaussian ODE-based flow path, where energy gradients drive efficient, few-step (1–2 steps) sampling. It is the first method to embed gradient guidance directly into the flow matching paradigm, unifying generation quality and energy calibration in a single optimization objective. The jointly trained energy model enables post-generation RMSD-aware ranking and precise ground-state identification. On GEOM-QM9 and GEOM-Drugs benchmarks, EnFlow achieves state-of-the-art generation performance with minimal sampling steps, significantly reduces ground-state energy prediction error, and improves both RMSD-based and energy-based ranking accuracy.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Natural Language Processing: Generation
• Machine Learning: Deep Generative Models & Autoencoders

Application Category

• 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
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Generating low-energy conformer ensembles and identifying ground-state conformations from molecular graphs remain computationally demanding with physics-based pipelines. Current learning-based approaches often suffer from a fragmented paradigm: generative models capture diversity but lack reliable energy calibration, whereas deterministic predictors target a single structure and fail to represent ensemble variability. Here we present EnFlow, a unified framework that couples flow matching (FM) with an explicitly learned energy model through an energy-guided sampling scheme defined along a non-Gaussian FM path. By incorporating energy-gradient guidance during sampling, our method steers trajectories toward lower-energy regions, substantially improving conformational fidelity, particularly in the few-step regime. The learned energy function further enables efficient energy-based ranking of generated ensembles for accurate ground-state identification. Extensive experiments on GEOM-QM9 and GEOM-Drugs demonstrate that EnFlow simultaneously improves generation metrics with 1--2 ODE-steps and reduces ground-state prediction errors compared with state-of-the-art methods.

Problem

• Generating diverse low-energy molecular conformers from graphs
• Identifying accurate ground-state conformations efficiently
• Unifying generative diversity with reliable energy calibration

Innovation

• Energy-guided flow matching for conformer generation
• Learned energy model enables accurate ground-state identification
• Unified framework improves fidelity with few-step sampling