Stochastic Population Genetic Models of Chronic Pathogens

Lead Research Organisation: University of Oxford
Department Name: Statistics

Abstract

Understanding the evolution of parasites which cause chronic infections is complicated by the fact that the relevant population genetic processes (selection and demographic stochasticity) occur both within and among infected hosts and operate on comparable time scales. Whereas it may sometimes be possible to model acute infections as though they were instantaneous over evolutionary time scales, such simplifications will be unsuitable when modeling the evolution of time-dependent traits such as those controlled by contingency loci or which have limited expression over the course of multi-stage parasite life cycles. Here we propose a number of models making use of measure-valued stochastic processes which we believe capture some of the biological phenomena relevant to chronic infections yet remain mathematically tractable. We consider both epidemics, which we propose to model using branching Markov processes, and endemic infections, which can be studied using various rigorous and heuristic approximations to a class of measure-valuedgeneralizations of the Moran model. We then describe how these models can be used to study three concrete phenomena: (i) the evolution of virulence in finite populations; (ii) host heterogeneity and diversifying selection on HIV-1; and (iii) pseudogene accumulation and the evolution of the VSG antigen repertoire in African trypanosomes.
 
Description New mathematical models were developed that capture the key strategies employed by parasites in causing chronic infections. These models retained sufficient analytic tractability that the `genealogies' of samples from populations evolving under the resulting, rather complex, conditions could be described. For example, persistent `bottlenecks' caused by extinction and recolonisation of populations within hosts, and the fluctuating selection resulting from changes in conditions experienced by the pathogens within a host were both considered. The importance of this is that it is these genealogies, which capture how pathogens in a sample are related to one another, that are inferred from data. In addition, efficient means of simulating genealogies for populations subject to natural selection in which one knows the genetic types of individuals in the sample were identified. This work has application beyond the original scope of the project.
Exploitation Route The work is several steps away from exploitation outwith academia. There is much further work to be done, but a better understanding of these models could ultimately result in more robust inference from data on pathogens.