DNA was extracted from whole blood and genotyped on the Illumina Omni-Express chip. Previous articles on COPDGene genome-wide association analyses reported details of initial quality control (QC) of genotype data [13, 17–23]. GWAS QC steps also included principal components calculation for both AA and NHW groups separately to summarize and adjust for their genetic ancestry. The hidden Markov model (HMM) based PennCNV algorithm was used to estimate CNVs exploiting the log R ratios (LRRs) and B allele frequencies (BAFs) from Illumina’s Omni-Express probes [24]. After CNV calling and QC, there were 9,076 samples (6,187 NHWs; 2,889 AAs) available for analysis of polymorphic CNVs (S1 Fig). Polymorphic CNVs were identified within each racial group, and those CNVs with >1% frequency among all subjects passing QC within each racial group became the genetic markers of interest for this analysis. Details of the QC process and CNV calling steps are discussed in Begum et al. (2015) [25].
