3D Graph Contrastive Learning for Molecular Property Prediction

TL;DR

This paper introduces a small-scale 3D graph contrastive learning framework for molecular property prediction, effectively utilizing 3D structural information with limited data and resources, achieving state-of-the-art results.

Contribution

The paper presents a novel contrastive learning method that incorporates 3D molecular structures using limited data and small models, addressing resource and information utilization issues.

Findings

Achieved state-of-the-art performance on four regression datasets.

Demonstrated the importance of 3D structural information in molecular representation.

Effective learning with only 1,128 samples using a small model.

Abstract

Self-supervised learning (SSL) is a method that learns the data representation by utilizing supervision inherent in the data. This learning method is in the spotlight in the drug field, lacking annotated data due to time-consuming and expensive experiments. SSL using enormous unlabeled data has shown excellent performance for molecular property prediction, but a few issues exist. (1) Existing SSL models are large-scale; there is a limitation to implementing SSL where the computing resource is insufficient. (2) In most cases, they do not utilize 3D structural information for molecular representation learning. The activity of a drug is closely related to the structure of the drug molecule. Nevertheless, most current models do not use 3D information or use it partially. (3) Previous models that apply contrastive learning to molecules use the augmentation of permuting atoms and bonds. Therefore, molecules having different characteristics can be in the same positive samples. We propose a novel contrastive learning framework, small-scale 3D Graph Contrastive Learning (3DGCL) for molecular property prediction, to solve the above problems. 3DGCL learns the molecular representation by reflecting the molecule's structure through the pre-training process that does not change the semantics of the drug. Using only 1,128 samples for pre-train data and 1 million model parameters, we achieved the state-of-the-art or comparable performance in four regression benchmark datasets. Extensive experiments demonstrate that 3D structural information based on chemical knowledge is essential to molecular representation learning for property prediction.