Chemical-Reaction-Aware Molecule Representation Learning

TL;DR

This paper introduces a novel molecule representation learning method that leverages chemical reactions to improve embedding organization and generalization, achieving state-of-the-art results across multiple chemical prediction tasks.

Contribution

The proposed approach uniquely uses chemical reaction constraints to enhance molecule embeddings, independent of GNN architecture, leading to improved performance and generalization.

Findings

Achieves 17.4% Hit@1 gain in reaction prediction

Improves AUC by 2.3% in property prediction

Reduces RMSE by 18.5% in graph-edit-distance prediction

Abstract

Molecule representation learning (MRL) methods aim to embed molecules into a real vector space. However, existing SMILES-based (Simplified Molecular-Input Line-Entry System) or GNN-based (Graph Neural Networks) MRL methods either take SMILES strings as input that have difficulty in encoding molecule structure information, or over-emphasize the importance of GNN architectures but neglect their generalization ability. Here we propose using chemical reactions to assist learning molecule representation. The key idea of our approach is to preserve the equivalence of molecules with respect to chemical reactions in the embedding space, i.e., forcing the sum of reactant embeddings and the sum of product embeddings to be equal for each chemical equation. This constraint is proven effective to 1) keep the embedding space well-organized and 2) improve the generalization ability of molecule embeddings. Moreover, our model can use any GNN as the molecule encoder and is thus agnostic to GNN architectures. Experimental results demonstrate that our method achieves state-of-the-art performance in a variety of downstream tasks, e.g., 17.4% absolute Hit@1 gain in chemical reaction prediction, 2.3% absolute AUC gain in molecule property prediction, and 18.5% relative RMSE gain in graph-edit-distance prediction, respectively, over the best baseline method. The code is available at https://github.com/hwwang55/MolR.