Recovering the period of an epidemic from a viral phylogeny: Exploring the limits of modern phylogenetic analysis using disease modeling
The early stages of epidemics are often unobserved because they occur far from public health infrastructure. This problem is exacerbated for diseases spreading in wildlife populations where monitoring is often logistically difficult. Modern molecular techniques joined with advanced phylogenetic modeling have provided a window into past events. Other modeling efforts have show that phylogenetic analysis of samples taken over the course of an epidemic can reveal periodic disease dynamics. Here we report on model results that further expand the envelope of phenomena that can be recovered from a viral phylogeny. We use our model to simulate the dynamics of a raccoon rabies outbreak and consider under what conditions a one-time sample of infected hosts can be used to recover periodicity of an epidemic from a viral phylogeny. Previous analysis of raccoon rabies data indicates a characteristic 48 month cycle. Our model simulates the spatial dynamics of raccoon rabies using a traditional SEI-type model solved using Gillespie’s numerical method. Our model includes the capacity to track the spread of disease between raccoons and the molecular evolution of the virus that occurs during the epidemic. We use skyline plots to analyze the viral phylogenies produced by our model.
We find that in principle, the periodicity of an epidemic can be recovered from a viral phylogeny based on a one-time sample of infected hosts. The population dynamics produced by our model exhibit a 48 month cycle. A skyline analysis of the simulated viral evolution that occurred during the epidemic also indicates a 48 month cycle in the effective number of infections (Ne,t). Furthermore, we find that changing the period of the epidemic, by changing demographic rates such as the transmission rate, produces a similar change in the period recovered from the viral phylogeny. The ability to recover the period of an epidemic is important because many infectious disease are not observed until well after their start. Knowledge of the epidemic period can be important in making plans for controlling a disease once it is discovered.