Phylogenetic Clustering of HIV and its Connection to Sexual Contact Networks
Phylogenetic models aim to estimate the ancestral relationships between DNA sequences. In doing so, they can help in the detection of HIV transmission clusters. What phylogenies reveal about the sexual contact network underlying an HIV epidemic is still unclear. The present work explores the link between phylogenetic clusters and sexual contact networks.
We simulate contact network structures and HIV transmission over these networks. Nodes remain infectious until diagnostic. We represent epidemics with phylogenies, from which we obtain simulated DNA samples. On a small set of epidemics, we partition DNA samples with both a distance-based and a numerically-intensive, maximum likelihood clustering algorithm. We compare the algorithms' clustering accuracy and, observing a limited gain from the maximum likelihood approach, we cluster the remaining samples with the distance-based approach. We graph several cluster characteristics against networks parameters, leading to the conclusion that network structure is reflected in the inferred phylogenetic clusters.