MUTATE: A Human Genetic Atlas of Multi-organ AI Endophenotypes using GWAS Summary Statistics

Artificial intelligence (AI) has been increasingly integrated into imaging genetics to provide intermediate phenotypes (i.e., endophenotypes) that bridge the genetics and clinical manifestations of human disease. However, the genetic architecture of these AI endophenotypes remains largely unexplored in the context of human multi-organ system diseases. Using publicly available GWAS summary statistics from the UK Biobank, FinnGen, and the Psychiatric Genomics Consortium, we comprehensively depicted the genetic architecture of 2024 multi- organ AI endophenotypes (MAEs). Two AI- and imaging-derived subtypes 1 showed lower polygenicity and weaker negative selection effects than schizophrenia disease diagnoses 2 , supporting the endophenotype hypothesis 3 . Genetic correlation and Mendelian randomization analyses reveal both within-organ relationships and cross-organ interconnections. Bi-directional causal relationships were established between chronic human diseases and MAEs across multiple organ systems, including Alzheimer’s disease for the brain, diabetes for the metabolic system, asthma for the pulmonary system, and hypertension for the cardiovascular system. Finally, we derived polygenic risk scores for the 2024 MAEs for individuals not used to calculate MAEs and returned these to the UK Biobank. Our findings underscore the promise of the MAEs as new instruments to ameliorate overall human health. All results are encapsulated into the MUTATE genetic atlas and are publicly available at https://labs-laboratory.com/mutate . Graphical abstract Key points Two AI- and neuroimaging-derived subtypes of schizophrenia (MAE-SCZ1 and MAE- SCZ2) show lower polygenicity and weaker negative selection signatures than the disease endpoint/diagnosis of schizophrenia, supporting the endophenotype hypothesis. Brain AI endophenotypes are more polygenic than other organ systems. Most multi-organ AI endophenotypes exhibit negative selection signatures, whereas a small proportion of brain patterns of structural covariance networks exhibit positive selection signatures. The 2024 multi-organ AI endophenotypes are genetically and causally associated with within-organ and cross-organ disease endpoints/diagnoses.