Use this URL to cite or link to this record in EThOS:
Title: Bayesian statistical modelling of genetic sequence evolution
Author: Angelis, K.
ISNI:       0000 0004 8503 1948
Awarding Body: UCL (University College London)
Current Institution: University College London (University of London)
Date of Award: 2016
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Bayesian statistics has been at the heart of phylogenetic inference over the last decade, particularly after the development of powerful programs that implement efficient Markov chain Monte Carlo algorithms, allowing inference from multi-parametric problems in realistic time frames. In this thesis we develop and test Bayesian methods to analyse molecular sequence data to address important biological questions. First, we review some fundamental aspects of Bayesian inference and highlight current Bayesian applications in molecular evolution with particular focus in studying natural selection and estimating species divergence times. Then, we develop a new Bayesian method to estimate the nonsynonymous/synonymous rate ratio and evolutionary distance for pairwise sequence comparisons. The new method addresses weaknesses of previous counting and maximum-likelihood methods. It is also computationally efficient and thus suitable for genome-scale screening. Then, we explore the performance of existing Bayesian algorithms in estimating species divergence times. In particular, we study the impact of ancestral population size and incomplete lineage sorting on Bayesian estimates of species divergence times under the molecular clock, when those factors of molecular evolution are ignored by the inference model. The estimates can be highly biased, especially in the case of shallow phylogenies with large ancestral population sizes. Then, using computer simulations and real data analyses we study the effect of five commonly used partitioning strategies for divergence times estimation and show that the choice of the partitioning scheme is important in case of serious clock-violation with incorrect prior assumptions. Finally, a Bayesian molecular clock dating study is performed to estimate the timeline of animal evolution. The results indicate that the time estimates are highly variable, precluding the inference of a precise timescale of animal evolution based on the current data and methods.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available