Comparison of the Multivariate Genetic Architecture of Eight Major Psychiatric Disorders Across Sex

Differences in the patterning of genetic sharing and differentiation between groups may arise from differences in biological pathways, social mechanisms, phenotyping and ascertainment. We expand the Genomic Structural Equation Modeling framework to allow for testing Genomic Structural Invariance (GSI): the formal comparison of multivariate genetic architecture across groups of individuals. We apply GSI to systematically compare the autosomal multivariate genetic architecture of eight psychiatric disorders spanning three broad factors (psychotic, neurodevelopmental, and internalizing) between cisgender males and females. We find that the genetic factor structure is largely similar between males and females, permitting meaningful comparisons of associations at the level of broad factors. However, problematic alcohol use loads on psychotic disorders in males but not in females, and both problematic alcohol use and post-traumatic stress disorder load more strongly on internalizing disorders in females than in males. Despite a high between-sex genetic correlation, the neurodevelopmental disorders factor exhibited weaker genetic correlations with psychotic and internalizing factors in females compared to males. Four biobehavioral phenotypes (educational attainment, insomnia, smoking ever, and Townsend Deprivation Index) had significant albeit small sex-differentiated associations with the psychotic factor. As GWAS samples continue to grow and diversify, GSI will become increasingly valuable for studying multivariate genetic architecture across groups.