Extracting and exploiting signals in genetic sequences : a thesis presented in partial fulfilment of the requirements for the degree of Doctor of Philosophy in Mathematics at Massey University
Year: 2011
Type: dissertation
Abstract: As DNA databases continue to grow at an exponential rate, the need for more efficient solutions to basic problems in computational biology grows ever more pressing. \nThese problems range from the principal questions driving evolutionary science: \nHow can we accurately infer the history of genes, individuals and species? How \ncan we separate the signal from the noise in our data? How can we visualise that \nsignal? - to the purely practical: How can we efficiently store all this data? With \nthese goals in mind, this thesis mounts a computational combination attack on a \nvariety of topics in bioinformatics and phylogenetics: \nA program is designed and implemented for solving the Maximum Parsimony \nproblem - in essence, finding phylogenetic trees having the fewest mutations. \nThis program generally outperforms existing highly optimised programs when \nusing a single CPU, and unlike these earlier programs, offers highly efficient \nparallelisation across multiple CPUs for further speedup.; \nA program is designed and implemented for compressing databases of DNA \nsequences. This program outperforms general-purpose compression by taking \nadvantage of the special "treelike" structure of DNA databases, using a novel \ndata structure, the "leaky move-to-front hashtable", to achieve speed gains.; \nA data visualisation technique is introduced that concisely summarises the \n"treelikeness" of phylogenetic datasets on a ternary plot. Each dataset is represented by a single point, allowing multiple datasets, or multiple treatments \nof a dataset, to be displayed on a single diagram.; \nWe demonstrate problems with a standard phylogenetic analysis methodology \nin which a single tree is assumed a priori. We argue for a shift towards network \nmethods that can in principle reject the hypothesis of a single tree.; \nMotivated by a phylogenetic problem, a fast new algorithm is developed for finding the mode(s) of a multinomial distribution, and an exact analysis of its \ncomplexity is given.
Author: Walton Timothy James White
Institution: Massey University
Language: English
Subfield: Molecular Biology
Domain: Life Sciences
Open Access status: green
Funders: University of Canterbury, Marsden Fund
OpenAlex ID: W2278281679
