DrugMCTS: a drug repurposing framework combining multi-agent, RAG and Monte Carlo Tree Search

2025-07-10

AI Summary

Large language models (LLMs) exhibit weak generalization and limited reasoning capability over unseen biomedical knowledge in drug repurposing, while fine-tuning and conventional retrieval-augmented generation (RAG) incur high computational overhead and neglect structured scientific data. Method: We propose DrugMCTS—a fine-tuning-free, multi-agent Monte Carlo Tree Search (MCTS) framework that integrates RAG, collaborative multi-agent reasoning, and MCTS to enable iterative, interpretable, and structured scientific reasoning. Contribution/Results: Built upon Qwen2.5-7B-Instruct, DrugMCTS achieves substantial improvements over generic LLMs and deep learning baselines on DrugBank and KIBA benchmarks, with over 20% higher recall. It demonstrates superior robustness, cross-domain generalization, and domain-specific adaptability—without requiring parameter updates or extensive retraining.

Technology Category

• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Multiagent Systems: Other Foundations of Multi Agent Systems
• Search and Optimization: Metareasoning and Metaheuristics

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Recent advances in large language models have demonstrated considerable potential in scientific domains such as drug discovery. However, their effectiveness remains constrained when reasoning extends beyond the knowledge acquired during pretraining. Conventional approaches, such as fine-tuning or retrieval-augmented generation, face limitations in either imposing high computational overhead or failing to fully exploit structured scientific data. To overcome these challenges, we propose DrugMCTS, a novel framework that synergistically integrates RAG, multi-agent collaboration, and Monte Carlo Tree Search for drug repurposing. The framework employs five specialized agents tasked with retrieving and analyzing molecular and protein information, thereby enabling structured and iterative reasoning. Without requiring domain-specific fine-tuning, DrugMCTS empowers Qwen2.5-7B-Instruct to outperform Deepseek-R1 by over 20%. Extensive experiments on the DrugBank and KIBA datasets demonstrate that DrugMCTS achieves substantially higher recall and robustness compared to both general-purpose LLMs and deep learning baselines. Our results highlight the importance of structured reasoning, agent-based collaboration, and feedback-driven search mechanisms in advancing LLM applications for drug discovery.

Problem

• Enhancing drug repurposing via multi-agent collaboration and structured reasoning
• Overcoming LLM limitations in scientific reasoning without fine-tuning
• Improving recall and robustness in drug discovery using feedback-driven search

Innovation

• Combines RAG, multi-agent, Monte Carlo Tree Search
• Employs five specialized agents for structured reasoning
• Enables feedback-driven search without fine-tuning