Major international projects are now underway aimed at creating a comprehensive catalog of all genes responsible for the initiation and progression of cancer. These studies involve sequencing of matched tumor–normal samples followed by mathematical analysis to identify those genes in which mutations occur more frequently than expected by random chance. Here, we describe a fundamental problem with cancer genome studies: as the sample size increases, the list of putatively significant genes produced by current analytical methods burgeons into the hundreds. The list includes many implausible genes (such as those encoding olfactory receptors and the muscle protein titin), suggesting extensive false positive findings that overshadow true driver events. Here, we show that this problem stems largely from mutational heterogeneity and provide a novel analytical methodology, MutSigCV, for resolving the problem. We apply MutSigCV to exome sequences from 3,083 tumor-normal pairs and discover extraordinary variation in (i) mutation frequency and spectrum within cancer types, which shed light on mutational processes and disease etiology, and (ii) mutation frequency across the genome, which is strongly correlated with DNA replication timing and also with transcriptional activity. By incorporating mutational heterogeneity into the analyses, MutSigCV is able to eliminate most of the apparent artefactual findings and allow true cancer genes to rise to attention.

Prostate cancer is the second most common cancer in men worldwide and causes over 250,000 deaths each year1. Overtreatment of indolent disease also results in significant morbidity2. Common genetic alterations in prostate cancer include losses of NKX3.1 (8p21)3,4 and PTEN (10q23)5,6, gains of the androgen receptor gene (AR)7,8 and fusion of ETS-family transcription factor genes with androgen-responsive promoters9–11. Recurrent somatic base-pair substitutions are believed to be less contributory in prostate tumorigenesis12,13 but have not been systematically analyzed in large cohorts. Here we sequenced the exomes of 112 prostate tumor/normal pairs. Novel recurrent mutations were identified in multiple genes, including MED12 and FOXA1. SPOP was the most frequently mutated gene, with mutations involving the SPOP substrate binding cleft in 6–15% of tumors across multiple independent cohorts. SPOP-mutant prostate cancers lacked ETS rearrangements and exhibited a distinct pattern of genomic alterations. Thus, SPOP mutations may define a new molecular subtype of prostate cancer.