arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2604.07669cs.LGcs.AIcs.CE
收藏

基于LLM引导的动作空间的强化学习用于可合成先导优化

LLM-Guided Dynamic Action Spaces for Synthesizable Molecular Optimization

Emory University(埃默里大学);University of Oxford(牛津大学);Independent Researcher(独立研究者)

Tao Li, Kaiyuan Hou, Tuan Vinh, Fanglei Xue, Monika Raj, Zhichun Guo, Carl Yang

AI总结:

本文提出MolReAct框架,通过合成约束的动作空间和GRPO算法优化先导分子,提升性质同时保证合成可行性,实现高效分子生成。

摘要:

在药物发现中,先导优化需要在改善治疗性质的同时确保分子修饰对应可行的合成路线。现有方法或优先考虑性质得分而不强制合成可行性,或依赖昂贵的大型反应网络枚举。本文引入MolReAct框架,将先导优化建模为一个马尔可夫决策过程,动作空间由经过验证的反应模板定义。一个工具增强的LLM代理作为动态反应环境,调用专用化学分析工具识别反应位点和功能基团,并提出基于匹配模板的紧凑化学基础转换。通过组相对策略优化(GRPO)训练的专用策略模型在多步轨迹中选择这些约束动作以最大化长期Oracle奖励,SMILES基于的缓存机制将端到端优化时间减少约43%。在13个属性优化任务和一个基于结构的对接任务中,MolReAct实现了平均Top-10得分为0.571,是所有基线中最高的,14项任务中13项排名第一或第二,9项任务达到最佳样本效率。通过将每一步优化都基于经过验证的反应模板,MolReAct生成的分子不仅性质得到改善,每个分子还伴随明确的模板基础合成路径。

英文摘要:

Synthesizable molecular optimization seeks to improve target properties while ensuring that molecular modifications follow feasible synthetic pathways. Existing synthesis-aware methods typically rely on exploring a large space of candidate transformations defined by reaction templates and purchasable building blocks. This search becomes even more challenging when property improvement requires multiple reaction steps, as the space expands further along the pathway. To address this challenge, we introduce MolReAct, which reformulates molecular optimization as search over compact reaction spaces proposed by a tool-augmented large language model (LLM). At each step, the LLM combines its prior chemical knowledge with cheminformatics tools to identify a molecule-specific set of compatible reactions, preserving synthesizability while making multi-step optimization feasible. Given this compact action space, we further leverage Group Relative Policy Optimization (GRPO) with the terminal oracle reward to improve long-term decision-making over multiple reaction steps. Across diverse molecular optimization tasks, MolReAct achieves the highest Top-10 score on 11 of 14 tasks and the best sample efficiency on 12 of 14 tasks, outperforming existing baselines under limited oracle budgets. Beyond these gains, MolReAct also provides each optimized molecule with a template-grounded synthetic pathway.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑