Reasoning-Enhanced Large Language Models for Molecular Property Prediction

2025-10-11

AI Summary

Molecular property prediction suffers from limited interpretability, poor cross-task generalization, and insufficient chemical reasoning capabilities. To address these challenges, we propose MPPReasoner—the first multimodal large language model integrating molecular images and SMILES representations, built upon the Qwen2.5-VL-7B-Instruct architecture. Our method introduces a novel chemistry-principle-guided reinforcement learning reward mechanism (RLPGR), enabling verifiable, rule-constrained chemical reasoning path generation. Training proceeds in two stages: expert-knowledge-guided supervised fine-tuning followed by principle-oriented reinforcement learning. Evaluated on eight benchmark datasets, MPPReasoner achieves improvements of 7.91% (in-distribution) and 4.53% (out-of-distribution) over state-of-the-art baselines, demonstrating superior generalization and chemically sound reasoning.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Knowledge Representation and Reasoning: Qualitative Reasoning
• Computer Vision: Multi-modal Vision

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation

Abstract

Molecular property prediction is crucial for drug discovery and materials science, yet existing approaches suffer from limited interpretability, poor cross-task generalization, and lack of chemical reasoning capabilities. Traditional machine learning models struggle with task transferability, while specialized molecular language models provide little insight into their decision-making processes. To address these limitations, we propose extbf{MPPReasoner}, a multimodal large language model that incorporates chemical reasoning for molecular property prediction. Our approach, built upon Qwen2.5-VL-7B-Instruct, integrates molecular images with SMILES strings to enable comprehensive molecular understanding. We develop a two-stage training strategy: supervised fine-tuning (SFT) using 16,000 high-quality reasoning trajectories generated through expert knowledge and multiple teacher models, followed by Reinforcement Learning from Principle-Guided Rewards (RLPGR). RLPGR employs verifiable, rule-based rewards that systematically evaluate chemical principle application, molecular structure analysis, and logical consistency through computational verification. Extensive experiments across 8 datasets demonstrate significant performance improvements, with MPPReasoner outperforming the best baselines by 7.91% and 4.53% on in-distribution and out-of-distribution tasks respectively. MPPReasoner exhibits exceptional cross-task generalization and generates chemically sound reasoning paths that provide valuable insights into molecular property analysis, substantially enhancing both interpretability and practical utility for chemists. Code is available at https://anonymous.4open.science/r/MPPReasoner-12687.

Problem

• Enhancing interpretability in molecular property prediction
• Improving cross-task generalization for drug discovery
• Incorporating chemical reasoning into AI decision processes

Innovation

• Multimodal model integrates molecular images with SMILES strings
• Two-stage training combines supervised fine-tuning and reinforcement learning
• Rule-based rewards evaluate chemical principles and logical consistency