Every study provided us with their most recent version of their haplotypes in VCF format with one VCF for every autosome. For every cohort, bcftools (v0.2.0-rc12) was used to create an entire-autosome, SNP-only site list with alternate and total allele count information from these per-chromosome haplotypes. Multiallelic SNPs were broken into biallelics using ‘bcftools norm’. These per-cohort site lists were merged into a single file using an in-house Perl script that correctly merges alternate and total allele counts. We created site lists called MAC2 and MAC5 containing only sites with a minor allele count (MAC) across all studies of >= 2 and >=5, respectively, using bcftools. These sites lists contained 95,855,206 and 51,060,347 sites, respectively.
