Title: Assessing Darwinian Selection in Tumors From Molecular “Big Data”
Speaker: Thomas LaFramboise (Associate Professor, Department of Genetics, Case Western Reserve University)
Abstract: Technological advances over the last decade now allow researchers to query the human genome for DNA variants in a near-comprehensive manner. This has been both a blessing and a curse, as the ability to detect all potential functional variants brings with it an enormous number of inconsequential ones. Assessing each for phenotypic impact raises a substantial multiple-testing problem. The problem is particularly acute in cancer genomics, where a single tumor may harbor hundreds of thousands of sporadic mutations. In cancer, one potential solution is to take advantage of the fact that Darwinian selection plays a central role in tumor growth, with mutations conferring a selective advantage to the cell leading to clonal expansions. These expansions result in vast portions of the tumor harboring the selected-for mutations, leaving clues that may be mined from the genomic data. In this talk, we review current methodology that aims to detect statistical signals of Darwinian selection from genomic data. We describe some of our own published work in this area for nuclear genome DNA variants, and also touch on our more recent efforts in the less well-studied mitochondrial genome.