P259: Clinical evaluation of DiagAI for streamlined variant interpretation in whole genome sequencing testing on Genomics England patients

The expanding adoption of genomic sequencing for diagnosing hereditary diseases has substantially increased the interpretive burden on clinical laboratories, necessitating efficient approaches to maximize diagnostic yield while maintaining feasible resource allocation. We developed DiagAI, an integrative machine-learning framework designed to prioritize and rank causal variants in rare diseases. DiagAI combines the Universal Pathogenicity Predictor (UP2) - a machine-learning model trained to predict ACMG pathogenicity classifications - with PhenoGenius for genotype-phenotype correlation analysis, alongside curated expert features including inheritance patterns and variant quality metrics.