The expression data for GIH, LWK, MEX and MKK populations were normalized for admixture using a customized version of EIGENSTRAT which generates principal components on the basis of genetic data [24]. Expression values were adjusted for each population using ten primary axes of variation from that population's corresponding intra-population PCA of the set of whole genome SNP genotypes. This correction for admixture also corrects for relatedness among some of the individuals in a few of the population samples as has been described in [25]. These residual normalized expression values were used as input for the association analysis.
