First, we aim to develop computational methods for analyzing viral DNA next-generation sequencing data that allow for predicting HIV-1 drug resistance. Viral DNA isolated from PBMCs is heavily hypermutated; hence statistical methods are needed that can separate hypermutations from causal drug resistance mutations. Second, building on our previous work on V-pipe, we aim at creating a computational pipeline that supports fully automated and reproducible end-to-end analyses of viral next-generation sequencing data in a diagnostic setting. Third, we aim to validate DNA-based drug resistance testing on a large dataset from the Swiss HIV Cohort Study comprising data from over 2000 people living with HIV. We will assess the sensitivity of detecting drug resistance mutations and their predictive value for treatment failure.
The remarkable success of antiretroviral therapy in HIV-infected patients is jeopardized by transmitted or acquired drug resistance. Testing for drug resistance mutations to select optimal personalized drug combinations is routinely performed by sequencing of the viral RNA genomes of plasma viruses. However, there is a second compartment of HIV-1, namely viral DNA in peripheral blood mononuclear cells (PBMCs), particularly relevant for people with low or undetectable plasma viremia. Sequencing viral DNA opens new diagnostic opportunities, but it also comes with new data analysis challenges.
