Genome-wide analysis was conducted as a linear regression of the additive effect of each SNP on BD in PLINK (Purcell et al. 2007). All autosomal SNPs that passed basic quality controls were tested for association with BD, and 10 ancestry dimensions were included as covariates. Age and sex were accounted for in the estimation of the BD phenotype. The criterion for individual SNP significance was set at p<5×10−8.
