RAPID: Improved phylogenetic approaches for characterizing the epidemiological dynamics of COVID-19

Biological Sciences - Phylogenetic trees constructed from viral genomes sampled from patients contain information about the historical pattern of transmission and dispersal of infectious diseases. Mathematical models of evolution allow researchers to infer critical epidemiological parameters, such as the transmission rate, from the information encoded in phylogenetic trees. Because future predictions and policy decisions depend on these estimates, it is crucial that their accuracy and limitations are well understood. Recent findings suggest that many commonly used mathematical models of disease evolution may yield highly inaccurate parameter estimates and may severely underestimate the associated uncertainty, thus potentially leading to sub-optimal policy decisions that are either ineffective or needlessly disruptive. This project will clarify precisely what epidemiological insights can be reliably inferred from phylogenetic trees and will develop new approaches to robustly characterize the spatial and temporal spread of COVID-19. Further, the project will determine which environmental, biological and policy factors affect the spread of COVID-19 based on phylogenetic data. The computational methods developed will be shared as standalone open source R packages that can be used for analyzing and guiding decisions against ongoing and future epidemics.

This project will use mathematical and computational methods to examine how and to what extent transmission, recovery and detection rates as well as the basic reproduction ratio (R0) and pathogen prevalence could possibly be estimated from phylogenetic data. The project will focus on models commonly used in phylogenetic epidemiology, including birth-death-sampling models and coalescent models with variable rates through time. The project will investigate whether the likelihood function of these models can be brought into special forms that reveal which epidemiological scenarios can possibly be statistically distinguished, similarly to the researchers' recent work on macroevolutionary birth-death models. Further, the researchers will develop more robust measures to characterize pathogen population dynamics over space and time from phylogenetic data, than currently possible. To that end, they will simulate a large number of epidemiological models and search for quantities that are preserved across statistically indistinguishable scenarios. The researchers will focus on the SARS-CoV-2 clade (the virus causing COVID-19) and will examine the implications of their findings for COVID-19 spread dynamics, based on published genomic data. In particular, they will perform comparative analyses between SARS-CoV-2 sub-clades as well as over space and time, to determine how various environmental, biological and policy factors affect the spread of the virus.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.