Genotype data pre-processing and imputations
Quality control of genotyped SNV data was carried out at both subject- and SNV-level using Plink 1.9, an open-source whole-genome analysis toolset.(23) Details are included in Supplemental Data 1 . In brief, subjects who were missing more than 10% of genotypes were omitted, as were those with sex discrepancies (i.e. reported sex did not match sex as determined by genetic markers on the sex chromosomes), and related subjects as determined using identity by descent measures. The processed genotype data was then imputed using the Michigan imputation server.(24) First, chromosomes were phased using ShapeIT (v2.r790), then SNVs were imputed using the Minimac algorithm and the Haplotype Reference Consortium reference panel (r1.1 2016).(25–27) SNVs that passed the quality score R2>0.3 were retained for further analysis.