We evaluated whether an excess of common SNPs with large allele frequency differences between Icelanders and Scots could provide evidence of population-specific natural selection. We used Icelanders and Scots (rather than Norwegians) in this analysis, because these samples were genotyped on the same chip under identical assay conditions, thus avoiding the effects of differential bias [26]. Indeed, tail distributions of comparisons between populations genotyped on different chips appear to be confounded by assay artifacts, precluding robust analyses of those comparisons (see Text S1). We used allele frequency differences between the Icelandic and Scottish samples at common SNPs to compute a χ2 (1 d.o.f.) statistic for unusual population differentiation that accounts for the effects of neutral genetic drift (see Materials and Methods). A P-P plot of our results is displayed in Figure 5. In contrast to Figure 4, there is a substantial excess of markers in the extreme tail, with 0.018% of markers having a P-value less than 0.0001. We speculate that many of these markers are likely to have been under natural selection.
