Speaker: Peter Chi, Associate Professor of Statistics, Department of Mathematics and Statistics, Villanova University
Title: Improving Distance-Based Phylogenetic Inference
Abstract: A phylogeny is typically thought of as an object that describes the evolutionary relationships between a group of species. Distance-based methods make up one class of procedures for phylogenetic inference, whereby pairwise evolutionary distances are first calculated among all of the species in the dataset utilizing their genetic information (e.g. DNA sequences), and then phylogenetic reconstruction proceeds by searching for the phylogeny that best matches the pairwise distances. This is thus a two-step procedure, in which inference is made on a summary statistic of the data (the pairwise distances) rather than on the data itself (the DNA sequence data). Here, we propose an improvement to distance-based phylogenetic inference by utilizing "robust distances" (O'Brien et al., 2009) that are functions of the tree and sequence data simultaneously. We show that usage of these robust distances in place of conventional distances results in gains to both branch length and topology estimation of phylogenetic trees within the ordinary least squares framework of phylogenetic inference.
