Poster - #UCLAFSPH-14 - Keywords: viral phylodynamics Hamiltonian Monte Carlo

Hamiltonian Monte Carlo sampling for phylodynamic inference using the episodic birth-death model.

Yucai Shao1, Andrew F. Magee2, Xiang Ji3, Philippe Lemey4 and Marc A. Suchard1,2,5 1Department of Biostatistics, Jonathan and Karin Fielding School of Public Health, University of California, Los Angeles, United States, 2 Department of Biomathematics, David Geffen School of Medicine at UCLA, University of California, Los Angeles, United States, 3 Department of Mathematics, School of Science & Engineering, Tulane University, New Orleans, United States, 4 Department of Microbiology and Immunology, Rega Institute, KU Leuven, Leuven, Belgium, 5 Department of Human Genetics, David Geffen School of Medicine at UCLA, University of California, Los Angeles, United States
(1 Department of Biostatistics, Jonathan and Karin Fielding School of Public Health, University of California, Los Angeles, United States) (2 Department of Biomathematics, David Geffen School of Medicine at UCLA, University of California, Los Angeles, United States) (3 Department of Human Genetics, David Geffen School of Medicine at UCLA, University of California, Los Angeles, United States)

ABSTRACT:
In the field of molecular epidemiology, genetic sequences can be used to reconstruct the evolutionary relationships and understand the dynamics of pathogen transmission over time. One type of phylogenetic branching model is the birth death models which can be used to provide insights into the temporal variation of speciation and extinction rates. However, existing methods for exploring the posterior distribution of phylogenetic models still lack computational efficiency, especially for large models and high-dimensional data. To overcome this issue, gradient-based sampling methods are employed to simultaneously update all dimensions of the parameter space and efficiently explore the high probability regions of the posterior distribution.