Abstract:
|
Phylodynamics is an area at the intersection of phylogenetics and population genetics that aims at reconstructing population size trajectories based on genetic data. Phylodynamic method relies on the coalescent, a stochastic point process that generates genealogies connecting randomly sampled individuals from the population of interest. Current existing approaches use non-parametric Gaussian process to estimate the effective population trajectory. However, those models cannot give further inference on attributes of the epidemiology, eg. infection rate. In this paper, we propose a parametric bayesian framework that combines phylodynamics inference and stochastic epidemiological model. In our framework, the population trajectory is model by Susceptible-Infected-Recovered (SIR) model. We use the Linear Noise Approximation (LNA) approach to approximate the SIR dynamics with a multivariate Gaussian process and use Markov chain Monte Carlo (MCMC) method to obtain posterior samples for the disease transmission parameters and latent population trajectory. Furthermore, we apply a period parameterization that allows inference infectious disease dynamics over multiple seasons.
|