We tested for population differences in gene expression levels as recent studies have shown it to be considerable 16,28. To avoid potential age and batch effects occurring at the establishment of the cell lines, we used the trios to establish critical values of differentiation. This approach (see Supplementary Methods) results in a difference in median log2 values of 0.2 (16% difference in median expression levels). Using this threshold, we tested all pairs of populations, after pooling CHB and JPT into one population (ASN) and calculated the number of genes exceeding this threshold. In total, 5,359 genes exceeded the threshold in one or more of the three population pairs (Figure S1), with most of them different in only a single population pair as expected. If the number of genes expressed in lymphoblastoid cell lines (LCLs) is about 50% of the total (about 12,000 genes) then we estimate that the fraction of genes with significant gene expression differences between any two populations is between 17% and 29%. However, one needs to be cautious with such a result since we observed that the
