Given the sample sizes, the likely effect sizes and frequencies of causal variants, and the proportion of causal variants in a gene, do current exome sequencing studies have sufficient power to detect genes underlying complex phenotypes? The enthusiasm about exome sequencing studies stems, in part, from successful candidate gene sequencing studies, and so we sought to test whether exome sequencing would be expected to have sufficient power to detect genes discovered by the candidate gene approach. So far, no published candidate gene study reported p-values that would be significant on the background of the complete exome (Table 2). This is particularly striking because some candidate gene studies used much larger sample sizes (thousands of individuals) than ongoing exome sequencing studies (hundreds of individuals). This demonstrates that current exome sequencing studies are underpowered to detect genes with the allelic distribution and effect sizes similar to the published examples. Indeed, extrapolation of effect sizes and frequencies from published studies shows (Figure 3) that thousands of individuals are required to reach acceptable statistical power. This analysis is consistent with an earlier study based
