We used the ENCODE data to assess how well each sample set could serve as a SNP discovery resource for other populations. This is an important practical matter, because it determines the effectiveness of scanning multiple populations for variation discovery as compared to sampling more deeply in a single population. To estimate how informative SNPs discovered in population A were for those present in population B, we counted the fraction of variants found in a sample of 30 A individuals that were also seen in a sample of 30 B individuals. Our measure of informativeness was the ratio of this fraction to that observed for a second, non-overlapping sample of 30 A individuals (Fig. 2a).
