Activity Number:
|
421
- Advances in Bayesian Modeling and Inferential Methods
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, July 31, 2018 : 2:00 PM to 3:50 PM
|
Sponsor:
|
Section on Bayesian Statistical Science
|
Abstract #330965
|
|
Title:
|
Bayesian Nonparametric Models for Multivariate Processes in Phylodynamics Using Stochastic Differential Equations
|
Author(s):
|
James Faulkner* and Vladimir N. Minin
|
Companies:
|
University of Washington and University of California, Irvine
|
Keywords:
|
Gaussian process;
coalescent;
population estimation;
phylogenetics;
infectious disease
|
Abstract:
|
One of the goals of phylodynamics is to estimate effective population size trajectories from genetic sequence data. Models based on Gaussian processes (GP) are the current state-of-the art in Bayesian nonparametric phylodynamic inference. However, computational challenges of GPs in high dimensions often require discrete approximations with Gaussian Markov random fields (GMRF) on a grid. A recent development uses stochastic partial differential equations (SPDE) and GMRFs to approximate GPs in continuous space for log Gaussian Cox processes (LGCP). We adapt the SPDE method for LGCPs to work with the coalescent likelihood and extend the method to accommodate multivariate processes related to effective population size trajectories. We assess model performance and computational efficiency with simulated data and compare results to other contemporary methods. We also apply the model to two multivariate data examples. The first example accounts for the sampling process for human influenza virus by jointly modeling it with the virus population size trajectory. The second example models an Ebola outbreak using a compartmental susceptible-infected-recovered model.
|
Authors who are presenting talks have a * after their name.