Buffy coat DNA from donors was extracted using the QIAsymphony SP system (QIAGEN) and submitted for genotyping on the Infinium Human CoreExome-24 BeadChip (Illumina) at the Center for Applied Genomics of the Children’s Hospital of Philadelphia. After filtering for missing genotypes (>5%) and deviation from Hardy-Weinberg equilibrium, haplotypes were estimated by statistical phasing using SHAPEIT2 (Delaneau et al., 2011). Missing genotypes were imputed with IMPUTE2 (Howie et al., 2009) using the 1000 Genomes Project multi-ethnic panel (Phase 3). Post-imputation SNPs with a missingness rate >5% or deviation from the Hardy-Weinberg equilibrium with P < 1 × 10−6 were removed. SNPs that had minor allele frequency ≥ 5% remained in eQTL analyses, resulting in a total of 5,313,398 autosomal SNPs. We estimated the population substructure among individuals used in our study by principal component analysis (PCA) of genotypes using smartpca as implemented in the EIGENSOFT suite (Price et al., 2006). A combined dataset containing all 90 individuals with one or both cell types represented in our study was used to estimate PCs. Resulting genotype PC estimates were also used as covariates for RNA-seq analysis.
