Digenic variant interpretation with hypothesis-driven explainable AI

Motivation: The digenic inheritance hypothesis holds the potential to enhance diagnostic yield in rare diseases. Computational approaches capable of accurately interpreting and prioritizing digenic combinations based on the proband's phenotypic profiles and familial information can provide valuable assistance to clinicians during the diagnostic process. Results: We have developed diVas, a hypothesis-driven machine learning approach that can effectively interpret genomic variants across different gene pairs. DiVas demonstrates strong performance both in classifying and prioritizing causative pairs, consistently placing them within the top positions across 11 real cases (achieving 73% sensitivity and a median ranking of 3). Additionally, diVas exploits Explainable Artificial Intelligence (XAI) to dissect the digenic disease mechanism for predicted positive pairs. Availability and Implementation: Prediction results of the diVas method on a high-confidence, comprehensive, manually curated dataset of known digenic combinations are available at oliver.engenome.com. ### Competing Interest Statement All the authors collaborate with enGenome srl. FDP, IL and SZ are full employees of enGenome srl. RB, PM, ER, IL and SZ have shares of enGenome.