ToolMol: Evolutionary Agentic Framework for Multi-objective Drug Discovery

2026-05-12

AI Summary

This work proposes ToolMol, a novel framework that integrates multi-objective genetic algorithms with tool-augmented large language model (LLM) agents to overcome the limitations of existing LLM-driven molecular generation methods, which often produce invalid or low-quality ligands due to syntactic constraints in molecular string representations. By leveraging precise chemical operations from the RDKit toolkit and guided by chain-of-thought reasoning, ToolMol iteratively optimizes ligand populations for efficient and synthesizable de novo small-molecule design. Experimental results across three protein targets demonstrate that ToolMol generates candidates with over 10% improvement in binding affinity and achieves absolute binding free energy scores more than 35% better than current state-of-the-art methods, while maintaining high validity and structural diversity.

Abstract

Advances in large language models (LLMs) have recently opened new and promising avenues for small-molecule drug discovery. Yet existing LLM-based approaches for molecular generation often suffer from high rates of invalid and low-quality ligand candidates, a result of the syntactic limitations of current models with regard to molecular strings. In this paper, we introduce $\texttt{ToolMol}$, an evolutionary agentic framework for de novo drug design. $\texttt{ToolMol}$ combines a multi-objective genetic algorithm with an agentic LLM operator that iteratively updates the ligand population. We build a comprehensive toolbox of RDKit-backed functions that allows our agentic operator to consisently make precise ligand modifications. $\texttt{ToolMol}$ achieves state-of-the-art performance on multi-objective property optimization tasks, discovering drug-like and synthesizable ligands that have $>10\%$ stronger predicted binding affinity compared to existing methods, evaluated on three protein targets. $\texttt{ToolMol}$ ligands additionally achieve state-of-the-art results in gold-standard Absolute Binding Free Energy scores, gaining over existing methods by over $35\%$. By studying chain-of-thought reasoning traces, we observe that tool-calling enables the model to more faithfully execute its planned modifications, efficiently exploiting the strong chemical prior knowledge in LLMs.

Problem

• molecular generation
• invalid ligands
• low-quality candidates
• syntactic limitations
• drug discovery

Innovation

• evolutionary agentic framework
• multi-objective drug discovery
• LLM-based molecular design
• tool-augmented reasoning
• binding affinity optimization