ABSTRACT
Somatic mutations occur in all body tissues throughout life and are not present in the germline from which the individual developed. One very important role of computational analysis of tumor samples is to support medical interpretation by collecting all information about the somatic mutations identified in the sample for which a targeted therapy might be available. The most important strategy for characterizing somatic mutations in cancer is the comparison of DNA from a cancer sample and a matched normal sample from the same individual. The majority of somatic mutations have no phenotypic effect. As with DNA sequencing of germline variants, the most important somatic alterations are single nucleotide variants (SNVs), short insertions and deletions (indels), and larger structural variants. The chapter demonstrates variant calling for somatic SNVs and indels based on a tumor-normal sample pair using VarScan2. VarScan2 employs a heuristic, pileup-based approach towards variant calling that shows good performance in practice.
