Using phylogenetic analysis of large-scale Next-Gen HIV sequence datasets and computer simulation to assess the impact of high-risk populations on the

Project Title: Using phylogenetic analysis of large-scale Next-Gen HIV sequence datasets and computer simulation to assess the impact of high-risk populations on the ART rollout in Uganda.

Background
A huge global expansion of the availability of antiretroviral (ARV) therapy for HIV-positive individuals has occurred in the last 10 years and now over 15 million, out of a total of 37 million, are on treatment1. Effective treatment stops HIV transmission and already there are annually 10% fewer new infections than 15 years ago although there are ~20% more HIV+ people. UNAIDS seeks to reduce HIV transmission by 25% by 2020 by increasing the proportion on treatment to 90% of those known to be infected2. However, it has been assuming that all communities in the "generalised" heterosexual HIV epidemics of sub-Saharan Africa are similar in transmission rate and access to care and this is far from the case.
Uganda has a strong reputation for managing its HIV epidemic well, but it has more than 1.5 million HIV+ individuals. Although in the rural population only 5% are HIV+, around Lake Victoria there are populations associated with the fishing industry, on the lake shore, or on islands, where up to 25% are HIV+. If, by transmission between communities, these populations contribute substantially to the national epidemic, the effectiveness of the national programme for ARV rollout will be threatened. This project will integrate HIV sequence analysis with evolutionary and epidemiologically driven mathematical and simulation models in order to estimate transmission between communities and predict its impact.

To understand how much difference the new data and methods will make the project will exploit an existing computational simulation model of an HIV epidemic that has been developed in the Leigh Brown group. By analysing available data for key features of the population contact structure (network structure and dynamics, as informed by our prior knowledge of HIV epidemiology), the model will be adjusted to match the known features of the populations studied, and will output virus genomes that arise from the type of population that is described. As the route by which they evolved, the "transmission tree" is known in the simulation, this allows testing of how good the methods of analysis are, and identifies key assumptions that could influence decision-making in regards to disease control and therefore require refinement.

Aims
Analysis of the genome data will inform the development of models of population contact. Making use of Prof Kao's expertise in network analysis of dense datasets, the aim will be to identify the influence that contact structure or mixing has on the observed infection rate, and in particular the relationship between individual level patterns of transmission, and the community level observations. This will inform our understanding of the importance of reducing infection rates in these communities for infection rates in the general population. As that may require greater expenditure per person than elsewhere, it will be important to demonstrate the national impact to persuade the agencies responsible to focus most effectively. The results of this work will be highly relevant to everyone involved in the rollout of ARV therapy in sub-Saharan Africa.

Training Outcomes
By the conclusion of the project the student will be expert in computational biology including programming in one or more languages, statistical analysis of sequence data (likelihood-based and Bayesian), epidemiological modeling in complex populations and familiar with issues around delivery of intervention strategies for global health.