Evolutionary analysis of rapidly evolving RNA viruses

Abstract

Recent advances in sequencing technology and computing power mean that we are in an unprecedented position to analyse large viral sequence datasets using state-of-the-art methods, with the aim of better understanding pathogen evolution and epidemiology. This thesis concerns the evolutionary analysis of rapidly evolving RNA viruses, with a focus on avian influenza and the use of Bayesian methodologies which account for uncertainty in the evolutionary process. As avian influenza viruses present an epidemiological and economic threat on a global scale, knowledge of how they are circulating and evolving is of substantial public health importance. In the first part of this thesis I consider avian influenza viruses of haemagglutinin (HA) subtype H7 which, along with H5, is the only subtype for which highly pathogenic influenza has been found. I conduct a comprehensive phylogenetic analysis of available H7 HA sequences to reveal global evolutionary relationships, which can help to target influenza surveillance in birds and facilitate the early detection of potential pandemic strains. I provide evidence for the continued distinction between American and Eurasian sequences, and suggest that the most likely route for the introduction of highly pathogenic H5N1 avian influenza to North America would be through the smuggling of caged birds. I proceed to apply novel methods to better understand the evolution of avian influenza. Firstly, I use an extension of stochastic mutational mapping methods to estimate the dN/dS ratio of H7 HA on different neuraminidase (NA) subtype backgrounds. I find dN/dS to be higher on the N2 NA background than on N1, N3 or N7 NA backgrounds, due to differences in selective pressure. Secondly, I investigate reassortment, which generates novel influenza strains and precedes human influenza pandemics. The rate at which reassortment occurs has been difficult to assess, and I take a novel approach to quantifying reassortment across phylogenies using discrete trait mapping methods. I also use discrete trait mapping to investigate inter-subtype recombination in early HIV-1 in Kinshasa, the epicentre of the HIV-1 group M epidemic. In the final section of the thesis, I describe a method whereby epidemiological parameters may be inferred from viral sequence data isolated from different infected individuals in a population. To conclude, I discuss the findings of this thesis in the context of other evolutionary and epidemiological studies, suggest future directions for avian influenza research and highlight scenarios in which the methods described in this thesis might find further application.