Fitting Structural Equation Models to GWAS-Derived Genetic Covariance Matrices

Genome-wide association studies (GWASs) are rapidly identifying loci affecting multiple phenotypes. Moreover, using cross-trait versions of methods such as genomic-relatedness-based restricted maximum-likelihood (GREML) and... [ view full abstract ]

Genome-wide association studies (GWASs) are rapidly identifying loci affecting multiple phenotypes. Moreover, using cross-trait versions of methods such as genomic-relatedness-based restricted maximum-likelihood (GREML) and LD-score regression (LDSC) researchers have identified genetic correlations between diverse behavioral, cognitive, psychiatric, and medical traits. These analyses are suggestive of constellations of phenotypes affected by shared sources of genetic liability, but they do not permit the causes of the observed genetic correlations to be investigated systematically. Here we introduce Genomic Structural Equation Modeling (Genomic SEM), a new method for modeling the multivariate genetic architecture of constellations of traits. Genomic SEM formally models the genetic covariance structure of GWAS summary statistics from samples of varying and potentially unknown degrees of overlap. We validate key properties of Genomic SEM with simulation and illustrate the flexibility and utility of Genomic SEM with several analyses of real summary data.