AOT*: Efficient Synthesis Planning via LLM-Empowered AND-OR Tree Search

2025-09-25

AI Summary

Multi-step retrosynthetic planning faces efficiency bottlenecks due to exponential search-space growth and high inference costs of large language models (LLMs). To address this, we propose LLM-Augmented AND-OR Tree Search (LAOTS), the first framework that atomically maps complete, LLM-generated synthetic routes onto an AND-OR tree structure. LAOTS introduces a mathematically rigorous hierarchical reward mechanism and retrieval-augmented contextual engineering to enable efficient, interpretable route exploration. Evaluated on standard benchmarks—including USPTO and MIT—LAOTS achieves state-of-the-art performance: it matches or exceeds solution rates of existing LLM-based methods while reducing required iterations by 60–80% (i.e., 1/3–1/5), with especially pronounced gains for complex molecules. This substantially lowers computational overhead and improves chemical plausibility, offering both scalability and domain interpretability.

Technology Category

• Planning, Routing, and Scheduling: Planning with Language Models
• Search and Optimization: Learning to Search
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Retrosynthesis planning enables the discovery of viable synthetic routes for target molecules, playing a crucial role in domains like drug discovery and materials design. Multi-step retrosynthetic planning remains computationally challenging due to exponential search spaces and inference costs. While Large Language Models (LLMs) demonstrate chemical reasoning capabilities, their application to synthesis planning faces constraints on efficiency and cost. To address these challenges, we introduce AOT*, a framework that transforms retrosynthetic planning by integrating LLM-generated chemical synthesis pathways with systematic AND-OR tree search. To this end, AOT* atomically maps the generated complete synthesis routes onto AND-OR tree components, with a mathematically sound design of reward assignment strategy and retrieval-based context engineering, thus enabling LLMs to efficiently navigate in the chemical space. Experimental evaluation on multiple synthesis benchmarks demonstrates that AOT* achieves SOTA performance with significantly improved search efficiency. AOT* exhibits competitive solve rates using 3-5$ imes$ fewer iterations than existing LLM-based approaches, with the efficiency advantage becoming more pronounced on complex molecular targets.

Problem

• Addresses computational challenges in multi-step retrosynthetic planning
• Improves efficiency and cost constraints of LLM-based synthesis planning
• Navigates exponential search spaces for complex molecular targets

Innovation

• Integrates LLM pathways with AND-OR tree search
• Uses reward strategy and retrieval-based context engineering
• Achieves high efficiency with fewer search iterations