SOLAR-Eclipse software (http://solar-eclipse-genetics.org) (50), in conjunction with the R package solarius (51), which uses maximum likelihood variance decomposition methods, was used to estimate phenotypic heritability (h2, the fraction of phenotypic variance attributable to additive genetic factors), as well as genetic (ρg) and unique environmental (ρe) correlations (i.e., the fraction of the correlation between 2 phenotypes that is attributable to either additive genetic or individual-specific environmental factors, respectively) of GMV and alcohol consumption. Post hoc analyses (Supplement 1) assessed the contribution of shared environmental factors to phenotypic correlations (ρc). These analyses were conducted among the subset of related participants from the HCP (n = 804; 293 families, 115 monozygotic and 64 dizygotic twin pairs and 422 nontwin siblings, excluding singletons and half-siblings). Covariates were identical to those in neuroimaging analyses. To ensure normality of measurements and accuracy of estimated parameters, an inverse normal transformation was applied to all continuous traits and covariates prior to analyses.
