PhyloEpid

Inferring epidemiological parameters from summary statistics of phylogenetic trees


Phyloepid

With the advent of affordable sequencing techniques, phylogenies are routinely gene- rated in epidemiological studies on viruses. The question is, can these really tell us more about epidemiological processes than classical incidence data? Recent works in the field of phylodynamics lead to the development of epidemiological parameter inference by maximum likelihood. But these methods are essentially based on simple epidemio- logical models like birth-death model. Indeed, they are limited by the difficulty to compute de likelihood function for more complex models.

Approximate Bayesian Computation (ABC) perform parameter inference and bypass the compu- tation of the likelihood function. ABC methods are based on simulation and comparison between target data and simulated data using summary statistics. We created a flexible simulation system implementing an event-driven approach of construction of phylogenies from epidemiological models. It is based on the assumption that phylogenies are similar to transmission trees. Then we designed summary statistics that summarize the epidemiolo- gical information of phylogenies. Inference precision by our method is really close to the precision obtained by a likelihood method implemented in BEAST. This work shows that phylogenies of viral sequences and ABC can inform us on epidemiological parameters and is a first step towards the analysis of more detailed epidemiological scenarios.

We think that within-host interactions between parasites largely determine how virulence evolves and that experimental data support model predictions. We also think that it is possible to make sense out of the complexity inherent to multiple infections and that experimental evolution settings may provide the best opportunity to further our understanding of virulence evolution. Finally, it is necessary to take into account epidemiological feedback to understand ecological and evolutionary effects of multiple infections.