Ranking mutated genes by significance for cancer. The cancer gene ranking score (Rs), derived from information reported in the COSMIC database, is defined as Rs = log2(Nm*Nc), where Nm is a number of unique cancer-associated mutations reported in the gene, and Nc is a number of different cancer types with mutations in this gene. All analyzed 3629 genes were divided into four categories depending on presence or absence of predicted functional mutations and known association to cancer (gene is considered as cancer associated, if it is annotated as TS or OG, or it interacts with one or more of TS or OG). Cancer associated genes are enriched with predicted functional mutations (P < 10−20 in two-tail Fisher test) compared to genes with unknown cancer association. Using a reasonable cutoff, one nominates a list of 957 genes with significance for cancer (arrow). A gene is above the cut either because it is observed to be multiply mutated (Rs > 1, three or more mutations) or, for Rs = 1 (two mutations), if at least one of the mutations in the gene is predicted as functional. Detailed statistical information on mutated genes is in Supplementary Table SM2. The higher proportion of genes with at least one predicted functional mutation (orange or brown) in frequently mutated genes (peak at left) is not surprising—in fact, a fair number of these mutations have been functionally validated in the literature. A particularly interesting set of genes (998, bottom left) are those that (so far) have been observed just once (Rs = 0) but contain a mutation predicted to be functional. Such genes may be rare, but functionally significant, contributors to oncogenesis and are good candidates for experimental follow-up.