Inferring Epidemic Contact Structure from Phylogenetic Trees

Abstract

Contact structure is believed to have a large impact on epidemic spreading and consequently using networks to model such contact structure continues to gain interest in epidemiology. However, detailed knowledge of the exact contact structure underlying real epidemics is limited. Here we address the question whether the structure of the contact network leaves a detectable genetic fingerprint in the pathogen population. To this end we compare phylogenies generated by disease outbreaks in simulated populations with different types of contact networks. We find that the shape of these phylogenies strongly depends on contact structure. In particular, measures of tree imbalance allow us to quantify to what extent the contact structure underlying an epidemic deviates from a null model contact network and illustrate this in the case of random mixing. Using a phylogeny from the Swiss HIV epidemic, we show that this epidemic has a significantly more unbalanced tree than would be expected from random mixing. One of the recent key innovations in the epidemiology of infectious diseases was the incorporation of explicit contact structure (i.e. who can infect whom) into epidemiological models. Theoretical studies have generated a broad consensus in the field that knowledge of the contact network may help to greatly improve the control of the spread of epidemics. The key problem in the field, however, is that we lack knowledge regarding the actual contact structure underlying real epidemics. Much research is focused on trying to reconstruct actual contact networks in various ways (mobile phone usage data, electronic devices that measure physical proximity, patient interviews, etc). All of these approaches are highly labour intensive and are fraught with many difficulties. Here, we present a new approach which is based on readily available sequence data. Using the Swiss HIV epidemic as an example, we show that it displays strong indications of a underlying contact structure that strongly differs from random interactions, thus undercutting the assumption of random mixing which is commonly made in epidemiological models.