Exome sequencing data contain an abundance of rare coding variation and indicate that a large fraction of this variation is functional. Not only are there many more rare variants than common ones, but sequencing additional samples continues to uncover additional rare variants. In fact, as sample size increases, the number of observed variants grows much faster than predicted by the neutral model of constant population size41, 42 (Figure 1). This relative excess of rare variants can be, in part, attributed to recent population expansion43–45, but is also likely due to purifying selection. As a consequence, rare variation is enriched for evolutionarily deleterious, and thus functional, variants. Additionally, the proportion of non-synonymous variants is higher among rare than among common variants45. Finally, among rare variants, missense variants predicted46 to be damaging are more prevalent than variants predicted to be benign (Figure 1). These findings are consistent with studies that demonstrated that rare variants in protein-coding regions are under purifying selection35, 47–51. Because sequencing larger samples continuously uncovers functionally relevant variants, exome sequencing studies enable direct identification of causal variants (in contrast to GWAS that use linkage-disequilibrium patterns between common markers).
