Rethinking Molecule Synthesizability with Chain-of-Reaction

2025-09-19

AI Summary

Existing molecular generation models struggle to guarantee synthetic accessibility and suffer from limited coverage of synthetically viable chemical space and suboptimal optimization performance. To address this, we propose Chain-of-Reaction (CoR) representation, which models synthesis pathways as stepwise reaction sequences—akin to reasoning paths in large language models—explicitly encoding chemical rules and enabling controllable, interpretable inference. Our framework integrates supervised learning for initialization, reinforcement learning for fine-tuning, and goal-directed test-time computation expansion to jointly model synthetic feasibility and guide optimization. Evaluated on synthesizable molecule reconstruction, molecular optimization, and hit expansion, our method achieves state-of-the-art performance: highest reconstruction accuracy, superior reaction-path diversity, and significantly improved optimization outcomes over all baselines.

Technology Category

• Search and Optimization: Learning to Search
• Constraint Satisfaction and Optimization: Constraint Learning and Acquisition
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

A well-known pitfall of molecular generative models is that they are not guaranteed to generate synthesizable molecules. There have been considerable attempts to address this problem, but given the exponentially large combinatorial space of synthesizable molecules, existing methods have shown limited coverage of the space and poor molecular optimization performance. To tackle these problems, we introduce ReaSyn, a generative framework for synthesizable projection where the model explores the neighborhood of given molecules in the synthesizable space by generating pathways that result in synthesizable analogs. To fully utilize the chemical knowledge contained in the synthetic pathways, we propose a novel perspective that views synthetic pathways akin to reasoning paths in large language models (LLMs). Specifically, inspired by chain-of-thought (CoT) reasoning in LLMs, we introduce the chain-of-reaction (CoR) notation that explicitly states reactants, reaction types, and intermediate products for each step in a pathway. With the CoR notation, ReaSyn can get dense supervision in every reaction step to explicitly learn chemical reaction rules during supervised training and perform step-by-step reasoning. In addition, to further enhance the reasoning capability of ReaSyn, we propose reinforcement learning (RL)-based finetuning and goal-directed test-time compute scaling tailored for synthesizable projection. ReaSyn achieves the highest reconstruction rate and pathway diversity in synthesizable molecule reconstruction and the highest optimization performance in synthesizable goal-directed molecular optimization, and significantly outperforms previous synthesizable projection methods in synthesizable hit expansion. These results highlight ReaSyn's superior ability to navigate combinatorially-large synthesizable chemical space.

Problem

• Addressing limited coverage of synthesizable molecule space
• Improving molecular optimization performance in generative models
• Enhancing step-by-step reasoning for chemical reaction pathways

Innovation

• Chain-of-reaction notation for stepwise chemical reasoning
• Reinforcement learning fine-tuning for enhanced reasoning capability
• Synthesizable molecule generation via pathway-based analog exploration