Speaker:Anindya Das

Date: Tue, 2017-10-24
Time: 3:10 pm
Location: Room 213, Atanasoff Hall

Title: Selection of Informative Genomic Regions for Closely Related Isolates
and Construction of their Phylogeny

Abstract: Building a phylogenetic tree for a number of species is a very
important part of understanding the evolutionary history of those species. If
some of these species are closely related and some are distantly related,
then the subtree(s) (or clade) of closely related isolates may have a very
low level of resolution. In these subtrees, sometimes there are some errors
in the phylogeny built by existing programs. The situation becomes worse when
different parts of the genomes of closely related species have multiple
contradictory histories of evolution. Then the subtrees for those closely
related species do not represent the representative phylogeny for any part of
the genome. To address these issues, we propose a method for identifying
low-resolution clades in the tree using branch lengths and likelihoods. We
introduce a novel approach for selecting informative regions of the genome
which capture the histories of evolution of these closely related species.
This approach is based on the likelihood deviations with respect to random
permutations of the isolates and nearest neighbor interchanges. We also
describe the technique of constructing representative and different (if
exist) phylogenies for different parts of the informative genomic regions for
the low-resolution clades. In support of these claims, we have found several
examples from SNP data of Fusarium virguliforme Isolates and sequences
generated using Monte Carlo simulation of molecular sequence evolution along
randomly generated phylogenetic trees.