With the aim of increasing the power of our cis- association analysis, data were combined (normalized expression values and SNP genotypes) for unrelated individuals of multiple populations to comprise three different multiple population analysis panels: 1) CEU-CHB-JPT-YRI, 2) CEU-CHB-JPT, and 3) CHB-JPT. The cis- association was performed separately for each of these panels using linear regression as described above, only considering those SNPs located less than 1Mb away from the probe midpoint. Conditional permutations were performed to assess significance of the nominal p-values 29,30.
