Investigations in phylogenetics: tree inference and model identifiability

Thumbnail
Name:
Yourdkhani_S_2020.pdf
Size:
2.635Mb
Format:
PDF
Download
Author
Yourdkhani, Samaneh
Chair
Rhodes, John A.
Allman, Elizabeth S.
Committee
McIntyre, Julie
Williams, Gordon
Keyword
phylogeny
mathematical models
evolutionary genetics
biology
computational biology
Metadata
Show full item record
URI
http://hdl.handle.net/11122/11303
Abstract
This thesis presents two projects in mathematical phylogenetics. The first presents a new, statistically consistent, fast method for inferring species trees from topological gene trees under the multispecies coalescent model. The algorithm of this method takes a collection of unrooted topological gene trees, computes a novel intertaxon distance from them, and outputs a metric species tree. The second establishes that numerical and non-numerical parameters of a specic Prole Mixture Model of protein sequence evolution are generically identifiable. Algebraic techniques are used, especially a theorem of Kruskal on tensor decomposition.
Description
Dissertation (Ph.D.) University of Alaska Fairbanks, 2020
Table of Contents
Chapter 1: Introduction. Chapter 2: Inferring Species Tree from Gene Trees -- 2.1 Introduction -- 2.2 Background and Notation -- 2.3 Weighted Rooted Triple Metrization of a Rooted Tree -- 2.4 Weighted Quartet Metrization of an Unrooted Tree -- 2.5 Weighted Quartet Distance Supertree and Consensus Algorithms -- 2.6 Algorithm Performance in Simulations -- References. Chapter 3: Identiability of a Protein Model -- 3.1 Abstract -- 3.2 Introduction -- 3.3 Markov Models on Trees -- 3.4 Algebraic Denitions and Lemmas -- 3.4.1 Denitions -- 3.4.2 Rank Propositions -- 3.5 Algebraic Aspects of the Prole Mixture Model -- 3.6 Identiability of Parameters for the Prole Mixture Model -- 3.7 Some other results -- References. Chapter 4: Conclusion and Future Work -- References -- Appendix.
Date
2020-05
Type
Dissertation
Collections