Analyzing combinations of somatic mutations in cancer genomes

In the past few years, high-throughput DNA sequencing has helped identify numerous genes that are recurrently mutated in cancer. Such recurrently mutated genes are likely to play key roles in the development of cancer. However, many other cancer genes are mutated rarely and therefore difficult to identify by their frequency of occurrence across cancer samples. Understanding the development and progression of cancer requires the identification of combinations of recurrently mutated genes in signaling and regulatory pathways. In this chapter, we discuss three approaches to identify such recurrently mutated combinations of genes: (1) evaluation of known pathways or gene sets, (2) discovery of significantly mutated subgraphs of an interaction network, and (3) identification of gene sets with mutually exclusive mutations. We demonstrate these three approaches on glioblastoma mutation data from the Cancer Genome Atlas.