Speaker
Description
Timed viral phylogenies are often used to understand geographic movements, past population dynamics, the emergence of new lineages and epidemic dynamics. These inferences are frequently done in a framework using coalescent theory. Exchangeability in coalescent theory refers to the property that each pair of lineages is equally likely to coalesce, moving back in time from the present to the past. Exchangeability forms part of the basis of phylodynamic likelihoods underpinning methods linking pathogen sequence data to epidemiological dynamics. However, exchangeability is a strong assumption, and while it is robust to many evolutionary and dynamic processes, there are also many processes that break it (selection, population structure, inhomogeneous sampling). Here we introduce a phylogeny-level statistic for exchangeability, and use it on simulated and real data. Our statistic can identify non-exchangeable dynamics in simulated phylogenies in which branching depends on the age of a lineage, and it can identify non-exchangeable dynamics of other origins. Meanwhile, our statistic for simulated phylogenies from neutral coalescent models indicates correctly that these are exchangeable. We explore combinations of this new statistic with phylogenetic signal to determine how well it may distinguish among different mechanisms that can create non-exchangeable patterns of ancestry. Finally, we apply these methods to SARS-CoV-2 sequence data, focusing on times in which VOC were emerging, to explore the interplay between competition, VOC population dynamics and exchangeability.
| Expedited Notification | No thanks, I do not require Expedited Notification |
|---|
