Graph Sampling-based Meta-Learning for Molecular Property Prediction

TL;DR

This paper introduces GS-Meta, a graph sampling-based meta-learning framework that leverages the many-to-many relationships between molecules and properties for improved few-shot molecular property prediction.

Contribution

The paper proposes a novel graph sampling approach within a meta-learning framework that models molecule-property relations as a graph and uses contrastive learning to enhance few-shot prediction.

Findings

GS-Meta outperforms state-of-the-art methods by 5.71%-6.93% in ROC-AUC.

The subgraph sampling strategy effectively captures topological information.

Contrastive loss improves the consistency and discrimination of subgraph episodes.

Abstract

Molecular property is usually observed with a limited number of samples, and researchers have considered property prediction as a few-shot problem. One important fact that has been ignored by prior works is that each molecule can be recorded with several different properties simultaneously. To effectively utilize many-to-many correlations of molecules and properties, we propose a Graph Sampling-based Meta-learning (GS-Meta) framework for few-shot molecular property prediction. First, we construct a Molecule-Property relation Graph (MPG): molecule and properties are nodes, while property labels decide edges. Then, to utilize the topological information of MPG, we reformulate an episode in meta-learning as a subgraph of the MPG, containing a target property node, molecule nodes, and auxiliary property nodes. Third, as episodes in the form of subgraphs are no longer independent of each other, we propose to schedule the subgraph sampling process with a contrastive loss function, which considers the consistency and discrimination of subgraphs. Extensive experiments on 5 commonly-used benchmarks show GS-Meta consistently outperforms state-of-the-art methods by 5.71%-6.93% in ROC-AUC and verify the effectiveness of each proposed module. Our code is available at https://github.com/HICAI-ZJU/GS-Meta.