FragFM: Efficient Fragment-Based Molecular Generation via Discrete Flow Matching

2025-02-19

AI Summary

To address low chemical validity, insufficient structural diversity, excessive sampling steps, and high computational complexity in molecular graph generation, this paper introduces the first fragment-based discrete flow matching framework. Methodologically, we propose a hierarchical autoencoder architecture that maps coarse-grained (fragment-level) representations to fine-grained (atom-level) ones, encoding molecules as invertible hierarchical graphs and performing discrete flow matching directly in fragment space to jointly model chemical constraints and structural priors. Compared with continuous-flow or autoregressive approaches, our framework reduces sampling steps by 10–50× while achieving >99% validity. Evaluated on ZINC, MOSES, and natural product benchmarks, the model maintains high structural diversity and precise property control, demonstrating superior efficiency, scalability, and chemical plausibility.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Search and Optimization: Mixed Discrete/Continuous Search
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Scalable techniques for the creation, curation, publication, maintenance, and consumption of large, Web-based, structured, reusable, knowledge graphs and ontologies

Abstract

We introduce FragFM, a novel fragment-based discrete flow matching framework for molecular graph generation.FragFM generates molecules at the fragment level, leveraging a coarse-to-fine autoencoding mechanism to reconstruct atom-level details. This approach reduces computational complexity while maintaining high chemical validity, enabling more efficient and scalable molecular generation. We benchmark FragFM against state-of-the-art diffusion- and flow-based models on standard molecular generation benchmarks and natural product datasets, demonstrating superior performance in validity, property control, and sampling efficiency. Notably, FragFM achieves over 99% validity with significantly fewer sampling steps, improving scalability while preserving molecular diversity. These results highlight the potential of fragment-based generative modeling for large-scale, property-aware molecular design, paving the way for more efficient exploration of chemical space.

Problem

• Fragment-based molecular generation
• Reduces computational complexity
• Improves scalability and validity

Innovation

• Fragment-based discrete flow matching
• Coarse-to-fine autoencoding mechanism
• High chemical validity with fewer steps