3D-Mol: A Novel Contrastive Learning Framework for Molecular Property Prediction with 3D Information

TL;DR

3D-Mol introduces a contrastive learning framework that enhances molecular property prediction by effectively capturing 3D spatial information and considering multiple conformations, outperforming existing methods on several benchmarks.

Contribution

It presents a hierarchical graph-based molecular encoding combined with contrastive pretraining on large unlabeled data, addressing limitations of previous models in spatial and conformational representation.

Findings

Outperforms state-of-the-art baselines on 7 benchmarks.

Effectively captures 3D spatial information and multiple conformations.

Improves molecular property prediction accuracy.

Abstract

Molecular property prediction, crucial for early drug candidate screening and optimization, has seen advancements with deep learning-based methods. While deep learning-based methods have advanced considerably, they often fall short in fully leveraging 3D spatial information. Specifically, current molecular encoding techniques tend to inadequately extract spatial information, leading to ambiguous representations where a single one might represent multiple distinct molecules. Moreover, existing molecular modeling methods focus predominantly on the most stable 3D conformations, neglecting other viable conformations present in reality. To address these issues, we propose 3D-Mol, a novel approach designed for more accurate spatial structure representation. It deconstructs molecules into three hierarchical graphs to better extract geometric information. Additionally, 3D-Mol leverages contrastive learning for pretraining on 20 million unlabeled data, treating their conformations with identical topological structures as weighted positive pairs and contrasting ones as negatives, based on the similarity of their 3D conformation descriptors and fingerprints. We compare 3D-Mol with various state-of-the-art baselines on 7 benchmarks and demonstrate our outstanding performance.