Co-mutation and exclusion analysis in human tumors, a tool for cancer biology studies and for rational selection of multitargeted therapeutic approaches.

Malignant tumors originate from somatic mutations and other genomic and epigenomic alterations, which lead to loss of control of the cellular circuitry. These alterations present patterns of co-occurrence and mutual exclusivity that can influence prognosis and modify response to drugs, highlighting the need for multi targeted therapies. Studies in this area have generally focused in particular malignancies and considered whole genes instead of specific mutations, ignoring the fact that different alterations in the same gene can have widely different effects. Here we present a comprehensive analysis of co-dependencies of individual somatic mutations in the whole spectrum of human tumors. Combining multi-testing with conditional and expected mutational probabilities, we have discovered rules governing the co-dependencies of driver and non-driver mutations. We also uncovered pairs and networks of co-mutations and exclusions, some of them restricted to certain cancer types and others widespread. These pairs and networks are not only of basic but also of clinical interest, and can be of help in the selection of multi targeted antitumor therapies. In this respect, recurrent driver co-mutations suggest combinations of drugs that might be effective in the clinical setting, while recurrent exclusions indicate combinations unlikely to be useful. This article is protected by copyright. All rights reserved.