Flow-Matching Based Refiner for Molecular Conformer Generation

2025-10-06

AI Summary

Low-energy molecular conformation generation (MCG) is critical in drug discovery, yet existing diffusion and flow-matching methods suffer from error accumulation during low signal-to-noise ratio (SNR) stages, degrading sampling quality. This paper proposes a flow-matching–based refinement framework that avoids low-SNR regimes via noise rescheduling and performs two-stage collaborative optimization over mixed-quality initial conformations generated by a denoising model. Key contributions include: (i) the first integration of flow matching into MCG refinement, decoupling initialization from refinement; (ii) an adaptive noise scheduling strategy that substantially reduces denoising steps; and (iii) simultaneous preservation of conformational diversity and improvement in geometric accuracy. Evaluated on GEOM-QM9 and GEOM-Drugs, our method achieves state-of-the-art performance across RMSD, Validity, and Diversity metrics, demonstrating superior efficiency and robustness over baseline approaches.

Technology Category

• Search and Optimization: Mixed Discrete/Continuous Search
• Machine Learning: Calibration & Uncertainty Quantification
• Planning, Routing, and Scheduling: Mixed Discrete/Continuous Planning

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG

Abstract

Low-energy molecular conformers generation (MCG) is a foundational yet challenging problem in drug discovery. Denoising-based methods include diffusion and flow-matching methods that learn mappings from a simple base distribution to the molecular conformer distribution. However, these approaches often suffer from error accumulation during sampling, especially in the low SNR steps, which are hard to train. To address these challenges, we propose a flow-matching refiner for the MCG task. The proposed method initializes sampling from mixed-quality outputs produced by upstream denoising models and reschedules the noise scale to bypass the low-SNR phase, thereby improving sample quality. On the GEOM-QM9 and GEOM-Drugs benchmark datasets, the generator-refiner pipeline improves quality with fewer total denoising steps while preserving diversity.

Problem

• Improving molecular conformer generation accuracy
• Reducing error accumulation in sampling process
• Optimizing noise scheduling for better performance

Innovation

• Flow-matching refiner initializes from mixed-quality outputs
• Reschedules noise scale to bypass low-SNR phase
• Improves sample quality with fewer denoising steps