Bio
Bienfait is a second-year PhD student in biomedical data science at the University of Oxford. He did his undergraduate studies in computer science at the University of Pennsylvania. As an undergraduate, he was a research assistant in bioinformatics. He was also briefly part of the Radiology AI lab at Brown University, contributing to the development of machine learning models that predict seizure using medical imaging data. Before starting his PhD in Oxford, he worked as a technology associate at Morgan Stanley, within the automated market making team. He is from Burundi.
Research interests
A prime goal in HIV vaccine research is to find antibodies that can bind to a wide range of HIV strains, also called broadly neutralizing antibodies (bnAbs). Combinations of bnAbs have been shown to hold the replication of HIV in the body at bay and hold great promises for HIV prevention, treatment and, if used in the right way, possibly even cure.
Successful bnAb combinations need to be able to neutralize most HIV strains from high-burden regions, such as sub-Saharan Africa. However, currently most bnAb research is lab-based, using HIV sequences isolated in Europe or the US. If bnAb performance could be predicted reliably from sequence data, it would open the possibility to include a wealth of African sequences generated over the last decade. Thus, Bienfait works on the development of reliable machine learning models to predict HIV’s resistance to bnAbs from sequence data. In particular, Bienfait is interested in applying natural language processing (NLP) techniques, since predicting bnAb resistance is fundamentally about predicting changes in interactions between proteins (i.e., HIV envelope protein and the antibodies), which can be modeled as sequences of amino acids, akin to sequences of characters in natural languages.
Beyond this specific project, Bienfait is generally interested in the development of interpretable machine learning models that can improve our knowledge about biological mechanisms.