MUTATE: a human genetic atlas of multiorgan artificial intelligence endophenotypes using genome-wide association summary statistics.

This article presents the MUTATE (MUlTiorgan AI endophenoTypE) genetic atlas, a comprehensive resource characterizing the genetic architecture of 2024 multiorgan artificial intelligence (AI) endophenotypes (MAEs) derived from imaging genetics data. Using publicly available genome-wide association study (GWAS) summary statistics from the UK Biobank, FinnGen, and the Psychiatric Genomics Consortium, the study estimates SNP-based heritability, polygenicity, natural selection signatures, genetic correlations, and causal relationships between MAEs and 525 disease endpoints (DEs) across multiple organ systems. Results reveal both organ-specific and cross-organ genetic correlations and bidirectional causal links, notably involving Alzheimer’s disease, diabetes, asthma, and hypertension. The study underscores the potential of AI-derived endophenotypes as intermediate traits bridging genetics and clinical manifestations, offering novel tools for understanding complex multiorgan diseases; all data and polygenic risk scores are publicly accessible via the MUTATE portal.