Modeling the evolution of drug resistance in HIV

Poster number: 0

Niko Beerenwinkel

  1. University of California, Department of Mathematics

The evolution of drug resistance in HIV is characterized by the accumulation of resistance-associated mutations in the viral genes coding for the drug targets. We model this basic evolutionary process with a class of probabilistic graphical models called mutagenetic trees. We show how mutagenetic trees and mixtures models of several trees can be learned from cross-sectional data. This correlation-based analysis can be significantly improved when longitudinal data is available. We present a hidden Markov model for learning the parameters of a mutagenetic tree from longitudinal clonal sequence data. We argue that mutagenetic trees capture many of the characteristics of interest of the development of HIV drug resistance, such as the order and the rate of accumulation of mutations. In particular, grouping patients according to different pathways to resistance or rates of progression may provide insight into diverse clinical outcome and provide the basis for individualized combination therapy.