Comment on "The consensus coding sequences of human breast and colorectal cancers"

Abstract

Sjöblom et al. (Research Articles, 13 October 2006, p. 268) used data from cancer genome resequencing to identify genes with elevated mutation rates. Their analysis used point probabilities when it should have used P values for the hypotheses they intended to test. Reimplementing their analysis method with exact P values results in far fewer genes with mutation rates that achieve statistical significance.