Bayesian birth-death skyline model : a case study on heterochronous Maltese SARS-Cov-2 genomic data

Abstract

When studying viral genome sequence data the Bayesian framework has the advantage that it can simultaneously construct phylogenetic trees and infer viral dynamics across time. This requires the specification of three models: (i) the transmission modelthe substitution model and the molecular clock model. In this study as transmission model we consider the Bayesian birth-death skyline (BDSKY) model and use the bModelTest method to define the substitution model. As a case study we consider 681 heterochronous genome sequences of COVID-19 sampled in Malta between 19/8/2020 and 5/1/2022. We consider both serial and multi-rho BDSKY models with two different molecular clock models: the strict and relaxed, and two settings for the number of intervals over which the reproductive number is considered constant (m=15 and m=30). In general the serial and the multi-rho BDSKY models gave considerably similar results yet some discrepancies were observed and these will be discussed. [excerpt]