Foundation Molecular Grammar: Multi-Modal Foundation Models Induce Interpretable Molecular Graph Languages

TL;DR

This paper introduces Foundation Molecular Grammar (FMG), a novel approach that uses multi-modal foundation models to induce interpretable molecular languages, improving molecular generation and property prediction with enhanced interpretability and efficiency.

Contribution

FMG leverages multi-modal foundation models to induce interpretable molecular grammars without expert annotations, enabling better molecular generation and discovery.

Findings

FMG improves synthesizability and diversity of generated molecules.

FMG enhances data efficiency in molecular tasks.

FMG provides built-in chemical interpretability.

Abstract

Recent data-efficient molecular generation approaches exploit graph grammars to introduce interpretability into the generative models. However, grammar learning therein relies on expert annotation or unreliable heuristics for algorithmic inference. We propose Foundation Molecular Grammar (FMG), which leverages multi-modal foundation models (MMFMs) to induce an interpretable molecular language. By exploiting the chemical knowledge of an MMFM, FMG renders molecules as images, describes them as text, and aligns information across modalities using prompt learning. FMG can be used as a drop-in replacement for the prior grammar learning approaches in molecular generation and property prediction. We show that FMG not only excels in synthesizability, diversity, and data efficiency but also offers built-in chemical interpretability for automated molecular discovery workflows. Code is available at https://github.com/shiningsunnyday/induction.