Genomic DNA from each sample was genotyped for approximately 710,000 SNPs using the Infinium OmniExpress-24 BeadChip array (Illumina). Genotyping data was subjected to stringent quality control using PLINK v1.9. and the check-bim utility from http://www.well.ox.ac.uk/~wrayner/tools/index.html. Samples with > 5% missing markers, ambiguous sex assignments, or anomalous heterozygosity were excluded. SNPs that were missing in > 5% of samples or had minor allele frequency less than 0.01 were removed, as were A/T and G/C SNPs with minor allele frequencies > 0.4. The SNP strand and ref/alt assignment were updated to match the Human Reference Consortium (HRC) version 1.1, and SNPs where the minor allele frequency differed by > 0.2 from the HRC were removed. Additional genotypes were imputed from the HRC panel using minimac3 and Eagle v2.3 phasing through the Michigan Imputation Server (https://imputationserver.sph.umich.edu/index.html). SNPs were annotated with rsID numbers from dbSNP v149 and converted to GRCh38 coordinates using CrossMap [58] with chain files from the UCSC Genome Browser (https://genome.ucsc.edu/). Non-SNP positions and duplicated sites identified by checkVCF (https://github.com/zhanxw/checkVCF) were filtered out, as were sites with an imputation r2 less than
