We restricted our attention to 7,966,132 autosomal markers (that is, SNPs) with MAF ≥ 0.01 and call rate ≥ 98%. In some analysis, severe linkage disequilibrium pruning was applied as follows: sliding windows of size 50 (that is, the number of markers used for linkage disequilibrium testing at a time) and window increments of 5 markers; for any pair of SNPs in a window, the first marker of the pair was discarded if r2 > 0.2. After linkage disequilibrium pruning, 1,089,227 SNPs were retained for analysis. All data-filtering procedures were conducted in PLINK v.1.964.
