An American-British team has developed a method for assessing how important mutations in specific noncoding regions of the genome are likely to be, and used it to look at nearly 100 cancer genomes to identify likely driver mutations in noncoding locations.