MolGuidance: Advanced Guidance Strategies for Conditional Molecular Generation with Flow Matching

2025-12-13

AI Summary

This work addresses conditional molecular generation, targeting the joint optimization of chemical validity, property alignment with target objectives, structural diversity, and sampling efficiency. We propose a novel guidance framework based on SE(3)-equivariant flow matching, introducing for the first time a decoupled guidance mechanism that separately handles continuous (velocity field) and discrete (logits) modalities. Guiding strength is jointly optimized across multiple strategies via Bayesian optimization. We systematically compare and rigorously characterize the applicability boundaries of classifier-free guidance, autoguidance, and model guidance in molecular generation—marking the first such analysis. Our method achieves new state-of-the-art property alignment on QM9 and QM9-14S, with chemical validity exceeding 98%, while significantly improving both sampling efficiency and structural diversity.

Technology Category

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

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Key objectives in conditional molecular generation include ensuring chemical validity, aligning generated molecules with target properties, promoting structural diversity, and enabling efficient sampling for discovery. Recent advances in computer vision introduced a range of new guidance strategies for generative models, many of which can be adapted to support these goals. In this work, we integrate state-of-the-art guidance methods -- including classifier-free guidance, autoguidance, and model guidance -- in a leading molecule generation framework built on an SE(3)-equivariant flow matching process. We propose a hybrid guidance strategy that separately guides continuous and discrete molecular modalities -- operating on velocity fields and predicted logits, respectively -- while jointly optimizing their guidance scales via Bayesian optimization. Our implementation, benchmarked on the QM9 and QMe14S datasets, achieves new state-of-the-art performance in property alignment for de novo molecular generation. The generated molecules also exhibit high structural validity. Furthermore, we systematically compare the strengths and limitations of various guidance methods, offering insights into their broader applicability.

Problem

• Develops hybrid guidance for conditional molecular generation
• Ensures chemical validity and property alignment in molecules
• Compares guidance methods for structural diversity and efficiency

Innovation

• Integrates classifier-free, autoguidance, and model guidance strategies
• Proposes hybrid guidance for continuous and discrete molecular modalities
• Jointly optimizes guidance scales via Bayesian optimization