Sex was included as a covariate (individuals whose self-reported gender was inconsistent with genetic sex were excluded from these analyses) in initial GWAS within both samples. For ALSPAC, wave at which the SRE items were first completed was included as a covariate; assessments are age-standardized but precise date of completion was unavailable. For S4S, age at which the SRE items were first completed was included as a covariate (mean [SD] age across all samples= 18.80 [0.79]). To account for population structure, 10 ancestry-informative principle components (PCs) were included in the ALSPAC GWAS, consistent with prior analyses using this sample (Edwards et al., 2017). S4S participants are of diverse ancestry, and were first assigned to empirically-based ancestry groups using principal components derived from 1000 Genomes (phase 3) reference populations, as described by Peterson and colleagues (Peterson et al., 2017). Subsequently, within-ancestry PCs were calculated to capture fine-grained stratification; PCs were retained as covariates in the GWAS using a stepwise regression approach.
