Title: Statistical approaches to unravel the life history of cancers
Authors: Stefan Dentro - Wellcome Trust Sanger Institute (United Kingdom) [presenting]
David Wedge - Big Data Institute Oxford (United Kingdom)
Peter Van Loo - The Francis Crick Institute (United Kingdom)
Abstract: Tumours evolve through the gradual acquisition of somatic mutations in the DNA of their cells. Some of these mutations provide a selective advantage to cell in which it occurred, giving rise to clonal expansions that allow the tumour to evolve and adapt. To better understand this process we have developed algorithms to disentangle the life history of tumours from a single or from multiple biopsy massively parallel sequencing data. These methods perform statistical inference over mutations and their properties, which are used as markers of tumour subpopulations that are in the process of expansion. Our algorithms are applied, as part of a larger effort, to the 2778 whole genome sequences in the International Cancer Genome Consortium Pan-Cancer Analysis of Whole Genomes project. We observe that nearly all tumours contain at least one distinct subpopulation, that the subpopulations contain mutations in known cancer genes and that these mutations are under positive selection. Furthermore, in the subpopulations we find mutations in genes relevant in the clinic. These findings suggest tumours continue to evolve up until diagnosis and may inform treatment choices.