The Rutgers University Cell and DNA repository genotyped saliva samples on the Smokescreen array (15). Genotyped calls were aligned to GRCh37 (hg19), and all individuals self-reporting ancestral origins (i.e., self-reported race) other than European or African American were excluded because these were the only ancestral populations in which GWAS summary statistics are available. Ancestrally homogenous samples were used due to differing genomic LD structure across ancestries and evidence that GWAS-derived summary statistics from one ancestral group cannot be meaningfully applied across ancestries (16,17). While novel techniques are being developed to leverage GWAS summary statistics across ancestries, these currently require availability of GWAS summary statistics from each ancestry. We separated individuals into self-reported ancestries prior to QC and conducted QC separately in the self-reported European and African ancestries subsamples. Following QC, genomic data were then used to exclude genomic ancestral outliers (described below).
