Using phylogeographic approaches to analyse the dispersal history, velocity and direction of viral lineages — Application to rabies virus spread in Iran

Abstract

Recent years have seen the extensive use of phylogeographic approaches to unveil the dispersal history of virus epidemics. Spatially explicit reconstructions of viral spread represent valuable sources of lineage movement data that can be exploited to investigate the impact of underlying environmental layers on the dispersal of pathogens. Here, we performed phylogeographic inference and applied different post hoc approaches to analyse a new and comprehensive data set of viral genomes to elucidate the dispersal history and dynamics of rabies virus (RABV) in Iran, which have remained largely unknown. We first analysed the association between environmental factors and variations in dispersal velocity among lineages. Second, we present, test and apply a new approach to study the link between environmental conditions and the dispersal direction of lineages. The statistical performance (power of detection, false‐positive rate) of this new method was assessed using simulations. We performed phylogeographic analyses of RABV genomes, allowing us to describe the large diversity of RABV in Iran and to confirm the cocirculation of several clades in the country. Overall, we estimate a relatively high lineage dispersal velocity, similar to previous estimates for dog rabies virus spread in northern Africa. Finally, we highlight a tendency for RABV lineages to spread in accessible areas associated with high human population density. Our analytical workflow illustrates how phylogeographic approaches can be used to investigate the impact of environmental factors on several aspects of viral dispersal dynamics.

Epidemiological hypothesis testing using a phylogeographic and phylodynamic framework

“…We further used the package "seraphim" to estimate two dispersal statistics based on the collection of such vectors: the mean lineage dispersal velocity (v mean ) and the weighted lineage dispersal velocity (v weighted ) 74 . While both metrics measure the dispersal velocity associated with phylogeny branches, the second version involves a weighting by branch time 75…”

“…Indeed, phylogeny branches with short duration have a lower impact on the weighted lineage dispersal velocity, which results in lower-variance estimates facilitating data set comparison 33 . On the other hand, estimating mean lineage dispersal velocity is useful when aiming to investigate the variability of lineage dispersal velocity within a distinct data set 75 . Finally, we also estimated the evolution of the maximal wavefront distance from the epidemic origin, as well as the evolution of the mean lineage dispersal velocity through time.…”

Phylogeographic and phylodynamic approaches to epidemiological hypothesis testing

“…To test if a particular environmental factor e tended to repulse viral lineages, BF e was then rather approximated with p e as the posterior probability that E estimated < E simulated . These tests are similar to a previous approach using a null dispersal model based on randomisation of phylogeny branches 73 .…”

“…We further used the package "seraphim" to estimate two dispersal statistics based on the collection of such vectors: the mean lineage dispersal velocity (v mean ) and the weighted lineage dispersal velocity (v weighted ) 72 . While both metrics measure the dispersal velocity associated with phylogeny branches, the second version involves a weighting by branch time 73 :…”

“…Indeed, phylogeny branches with short duration have a lower impact on the weighted lineage dispersal velocity, which results in lower-variance estimates facilitating data sets comparison 33 . On the opposite, estimating mean lineage dispersal velocity is rather useful when aiming to investigate the variability of lineage dispersal velocity within a distinct data set 73 . Finally, we also estimated the evolution of the maximal wavefront distance from the epidemic origin, as well as the evolution of the mean lineage dispersal velocity through time.…”