DNA from study participants was extracted from blood, saliva, or immortalized cell lines. Subjects were genotyped using either the Illumina HumanOmni1-Quad_v1.0 microarray, or the Illumina Human Core Exome microarray. Subjects were genotyped on the HumanOmni1-Quad_v1.0 at the Yale Center for Genome Analysis or the Center for Inherited Disease Research (CIDR). The HumanOmni1-Quad_v1.0 contains 988,306 autosomal SNPs, and genotypes were called using Illumina Genome Studio software v2011.1, genotyping module v1.8.4. Subjects were initially filtered based on call rate, and SNPs filtered based on call rate and frequency, with identity-by-descent (IBD) estimates used to quantify genetic relatedness between subjects. Extensive details on genotyping and data cleaning procedures have been published previously.24 Subjects genotyped on the HumanOmni1-Quad_v1.0 were included in the present study if they were unrelated and either self-reported African-Americans (AAs) or European-Americans (EAs), with population outliers then removed based on principal component analysis (PCA) of genotype data.31, 32 Subjects not genotyped on the HumanOmni1-Quad_v1.0 were genotyped on the Human Core Exome microarray, which contains both exome-focused SNP content and tagging SNPs for genome-wide imputation. Details about the application of quality control
