The large-scale distribution of somatic mutations in cancer genomes

Recently, the genome sequences from sev- eral cancers have been published, along with the genome from a noncancer tissue from the same individual, allowing the identification of new somatic mutations in the cancer. We show that there is significant variation in the den- sity of mutations at the 1-Mb scale within three cancer genomes and that the density of mutations is correlated between them. We also demonstrate that the density of mutations is correlated to that in the germline, as mea- sured by the divergence between humans and chimpanzees and humans and macaques. We show that the density of mutations is correlated to the guanine and cytosine (GC) conent, replication time, distance to telomere and cen- tromere, gene density, and nucleosome occupancy in the cancer genomes. However, overall, all factors explain less than 40% of the variance in mutation density and each factor explains very little of the variance. We find that genes associated with cancer occupy regions of the genome with significantly lower mutation rates than the average. Finally, we show that the density of mutations varies at a 10-Mb and a chromosomal scale, but that the variation at these scales is weak.