MolEvolve: LLM-Guided Evolutionary Search for Interpretable Molecular Optimization

TL;DR

MolEvolve introduces an LLM-guided evolutionary framework for molecular discovery that enhances interpretability and effectively addresses activity cliffs, outperforming traditional methods in property prediction and molecule optimization.

Contribution

It presents a novel autonomous, look-ahead planning approach using LLMs and MCTS to discover interpretable chemical transformations without relying on human-engineered features.

Findings

Outperforms baselines in molecule optimization tasks

Generates transparent, human-readable chemical insights

Effectively addresses activity cliffs in molecular property prediction

Abstract

Despite deep learning's success in chemistry, its impact is hindered by a lack of interpretability and an inability to resolve activity cliffs, where minor structural nuances trigger drastic property shifts. Current representation learning, bound by the similarity principle, often fails to capture these structural-activity discontinuities. To address this, we introduce MolEvolve, an evolutionary framework that reformulates molecular discovery as an autonomous, look-ahead planning problem. Unlike traditional methods that depend on human-engineered features or rigid prior knowledge, MolEvolve leverages a Large Language Model (LLM) to actively explore and evolve a library of executable chemical symbolic operations. By utilizing the LLM to cold start and an Monte Carlo Tree Search (MCTS) engine for test-time planning with external tools (e.g. RDKit), the system self-discovers optimal trajectories autonomously. This process evolves transparent reasoning chains that translate complex structural transformations into actionable, human-readable chemical insights. Experimental results demonstrate that MolEvolve's autonomous search not only evolves transparent, human-readable chemical insights, but also outperforms baselines in both property prediction and molecule optimization tasks.