Zero-shot drug repurposing with geometric deep learning and clinician centered design

Abstract

Historically, drug repurposing – identifying new therapeutic uses for approved drugs – has been attributed to serendipity. While recent advances have leveraged knowledge graphs and deep learning to identify potential therapeutic candidates, their clinical utility remains limited due to their dependence on existing knowledge about diseases. Here, we introduce TXGNN, a geometric deep learning approach designed for “zero-shot” drug repurposing, enabling therapeutic predictions even for diseases with no existing medicines. Trained on a medical knowledge graph, TXGNN utilizes a graph neural network and metric-learning module to rank therapeutic candidates as potential indications and contraindications across 17,080 diseases. When benchmarked against eight leading methods, TXGNN significantly improves prediction accuracy for indications by 49.2% and contraindications by 35.1% under stringent zero-shot evaluation. To facilitate interpretation and analysis of the model’s predictions, TXGNN’s Explainer module offers transparent insights into the multi-hop paths that form TXGNN’s predictive rationale. Clinicians and scientists found TXGNN’s explanations instrumental in contextualizing and validating its predicted therapeutic candidates during our user study. Many of TXGNN’s novel predictions have shown remarkable alignment with off-label prescriptions made by clinicians within a large healthcare system, affirming their real-world utility. TXGNN provides drug repurposing predictions that are more accurate than existing methods, consistent with off-label prescription decisions made by clinicians, and can be investigated through multi-hop interpretable explanations.