Mutation-Profile-Based Methods for Understanding Selection Forces in Cancer Somatic Mutations: A Comparative Analysis

Abstract

Human genes exhibit different effects on fitness in cancer and normal cells. Here, we present an evolutionary approach to measure the selection pressure on human genes, using the well-known ratio of the nonsynonymous to synonymous substitution rate in both cancer genomes (C N /C S ) and normal populations (p N /p S ). A new mutationprofile-based method that adopts sample-specific mutation rate profiles instead of conventional substitution models was developed. We found that cancer-specific selection pressure is quite different from the selection pressure at the species and population levels. Both the relaxation of purifying selection on passenger mutations and the positive selection of driver mutations may contribute to the increased C N /C S values of human genes in cancer genomes compared with the p N /p S values in human populations. The C N /C S values also contribute to the improved classification of cancer genes and a better understanding of the onco-functionalization of cancer genes during oncogenesis. The use of our computational pipeline to identify cancerspecific positively and negatively selected genes may provide useful information for understanding the evolution of cancers and identifying possible targets for therapeutic intervention.

Use of signals of positive and negative selection to distinguish cancer genes and passenger genes

“…We observed a greater similarity when we compared our list of negatively selected genes with that of Zhou et al [51]: 32 of the 112 genes identified by Zhou et al,[51] are also present among the 505 negatively selected genes identified in the present work (Supplementary file 3).…”

“…In summary, the pro-oncogenic, negatively selected genes annotated and validated in the present work are missing from the earlier lists of negatively selected genes [51,66,67,73]. A possible explanation for the lack of similarity of top-ranking negatively selected genes identified in the present study with those identified by others is that we have limited our work to transcripts that have at least 100 somatic mutations.…”

“…It is noteworthy in this respect that another member of the family of chemokine receptors, the atypical chemokine receptor 3, ACKR3 is also among the genes showing very high values of rSMN, suggesting negative selection of missense and nonsense mutations (Supplementary file 3). (Note that Zhou et al, [51] have also identified ACKR3 as a negatively selected gene). Significantly, ACKR3 is a well-known oncogene, present in Tier 1 of the Cancer Gene Census.…”

“…In harmony with some of these expectations, using just the ratio of the nonsynonymous to synonymous substitution rate as a measure of positive or negative selection, Zhou et al [51] have shown that in cancer genomes, the majority of genes had nonsynonymous to synonymous substitution rate values close to one, suggesting that they belong to the passenger gene category.…”

“…In a more recent study, Zhou et al [51] have identified 365 genes for which the ratio of the nonsynonymous to synonymous substitution rate was significantly increased, suggesting that they are subject to the positive selection of driver mutations. However, an obvious limitation of such approaches is that they implicitly assume that synonymous substitutions are selectively neutral and therefore the ratio of the nonsynonymous to synonymous substitution rate properly monitors selection.…”