HANAMI predicts drug-gene-disease motifs with heterogeneous graph contrastive learning, improving up to about 6% over state-of-the-art baselines and holding an ~18% edge in zero-shot settings
The authors present HANAMI (Heterogeneous grAph coNtrastive leArning for drug-gene-disease Motif predIction), a multi-view deep graph learning framework that integrates heterogeneous biomedical knowledge such as chemical structures, genomic sequences, and clinical phenotypes and uses relation-aware topology encoding, structure-aware aggregation, and contrastive learning to predict drug-gene-disease motifs; systematic evaluation on benchmark datasets shows up to about 6% improvement over existing state-of-the-art methods in motif prediction, an approximately 18% performance advantage maintained in zero-shot settings involving previously unseen entities, and the ability to prioritize drug-disease relationships investigated in Phase II or III trials while identifying candidate genes suggestin