Hierarchical Molecular Representation Learning via Fragment-Based Self-Supervised Embedding Prediction

TL;DR

This paper introduces GraSPNet, a hierarchical self-supervised learning framework that models atomic and fragment-level semantics in molecular graphs, leading to improved molecular property prediction.

Contribution

It proposes a novel hierarchical GSSL method that captures multi-resolution chemical structures without predefined vocabularies, enhancing transferability.

Findings

Outperforms state-of-the-art GSSL methods on molecular benchmarks

Learns chemically meaningful and transferable representations

Effectively models multi-level molecular semantics

Abstract

Graph self-supervised learning (GSSL) has demonstrated strong potential for generating expressive graph embeddings without the need for human annotations, making it particularly valuable in domains with high labeling costs such as molecular graph analysis. However, existing GSSL methods mostly focus on node- or edge-level information, often ignoring chemically relevant substructures which strongly influence molecular properties. In this work, we propose Graph Semantic Predictive Network (GraSPNet), a hierarchical self-supervised framework that explicitly models both atomic-level and fragment-level semantics. GraSPNet decomposes molecular graphs into chemically meaningful fragments without predefined vocabularies and learns node- and fragment-level representations through multi-level message passing with masked semantic prediction at both levels. This hierarchical semantic supervision enables GraSPNet to learn multi-resolution structural information that is both expressive and transferable. Extensive experiments on multiple molecular property prediction benchmarks demonstrate that GraSPNet learns chemically meaningful representations and consistently outperforms state-of-the-art GSSL methods in transfer learning settings.