Within all 49 cohorts we performed PCA using autosomal SNPs with high imputation quality (INFO >0.8), low missingness (<1%), MAF>0.05 and in relative linkage equilibrium after 2 iterations of linkage disequilibrium (LD) pruning (r2 < 0.2, 200 SNP windows), removing well known long-range-LD areas (MHC and chr8 inversion). 17,608 SNPs present in all 49 cohorts, followed by the above LD pruning were used for robust relatedness testing across cohorts using PLINK v1.955; pairs of subjects with PIHAT > 0.2 were identified and one member of each pair removed at random, preferentially retaining cases over controls.
