For the common SNP association analysis the following criteria were applied for exclusion of SNPs prior to imputation: (i) minor allele frequency (MAF)<0.01, (ii) Hardy-Weinberg equilibrium (HWE)p<10−4, (iii) call-rate<0.95 if MAF>0.05, call-rate<0.97 if 0.02<=MAF<=0.05 and call-rate<0.99 if 0.01<=MAF<0.02. Samples were excluded prior to imputation based on the following criteria: (i) call-rate<0.95, (ii) heterozygosity outside the population-specific bounds ((iii) relatedness – 500 SNPs with 0.45≤ minor allele frequency<0.5 and call-rate>0.95 and 1000000 bp apart were used to estimate the concordance between pairs of samples. If the genotype concordance for two individuals is >0.7 and <0.97 (related) or ≥0.97 (duplicate) then the one with the lowest call rate was excluded. (iv) gender checks: gender was inferred from the intensity of the A-allele probes on the X chromosome normalised against the autosomal intensities. Individuals for which gender could not be inferred were excluded. (v) intensity failure: individuals were excluded if the mean of their A and B allele intensities from 10000 SNPs on chromosome 22 was outlying when compared to the sample at large. (vi) identity: prior to being genotyped, ~30 SNPs were
