Molecular property prediction improved by combining graph and fingerprint views
Multi-View Molecular Representation Learning with Hierarchical Graphs and Contextualized Fingerprints
Machine LearningArtificial Intelligence
Summary
Predicting how molecules will behave or their properties is tough when there isn’t much data on new kinds of molecules. The researchers created a new system called HiFi-Mol that looks at molecules in two ways: one examines the molecule's structure like a map of connected atoms, and the other looks at chemical features called fingerprints in context. By training these two parts separately and then combining them, HiFi-Mol can better predict molecular properties, especially for molecules different from those seen before.
What this means in practice
- •For drug discovery teams: Enhance prediction of novel molecule properties to speed up identifying promising drug candidates with limited experimental data.
- •For chemical data scientists: Use combined graph and fingerprint representations to improve accuracy when modeling and predicting chemical compound activities.
Authors
Gwang-Hyeon Yun, Jong-Hoon Park, Bing Hu, Helen Chen, Anita Layton, Young-Rae Cho
Abstract
Molecular property prediction requires representations that generalize from limited labeled data to structurally novel compounds. Existing molecular pretraining methods often rely on a single view: graph-based approaches model atom-bond topology but provide limited fragment-level supervision, whereas fingerprint descriptors encode chemical patterns but are typically used as fixed auxiliary features. We propose HiFi-Mol, a multi-view framework that separately pretrains a hierarchical graph encoder and a contextualized fingerprint encoder before downstream integration. The graph branch uses fragment-aware masking with multi-resolution supervision to capture substructure-aware representations, while the fingerprint branch tokenizes active entries from seven fingerprint families and applies masked language modeling to learn contextualized embeddings. During fine-tuning, HiFi-Mol combines projected multi-resolution graph features with fingerprint embeddings for downstream prediction. Evaluated on MoleculeNet benchmarks under the scaffold split, HiFi-Mol achieves a 2.77% improvement in average ROC-AUC over the best baseline across eight classification tasks while maintaining competitive performance on three regression tasks. Further analyses reveal that fragment-aware masking improves graph representation quality, and classification results demonstrate dataset-dependent strengths of the individual graph and fingerprint variants, confirming that the two views provide complementary predictive signals.