Inferring intra-tumor heterogeneity from high-throughput DNA sequencing data

Cancer is a disease driven in part by somatic mutations that accumulate during the lifetime of an individual. The clonal theory [1] posits that the cancerous cells in a tumor are descended from a single founder cell and that descendants of this cell acquired multiple mutations beneficial for tumor growth through rounds of selection and clonal expansion. A tumor is thus a heterogeneous population of cells, with different subpopulations of cells containing both clonal mutations from the founder cell or early rounds of clonal expansion, and subclonal mutations that occurred after the most recent clonal expansion. Most cancer sequencing projects sequence a mixture of cells from a tumor sample including admixture by normal (non-cancerous) cells and different subpopulations of cancerous cells. In addition most solid tumors exhibit extensive aneuploidy and copy number aberrations. Intra-tumor heterogeneity and aneuploidy conspire to complicate analysis of somatic mutations in sequenced tumor samples.