Phylodynamics beyond neutrality: The impact ofincomplete purifying selection on viral phylogeniesand inference

Viral phylodynamics focuses on using sequence data to make inferences about the population dynamics of viral infectious diseases. These inferences commonly include estimation of the viral growth rate, the reproduction number, and the time of most recent common ancestor. With few exceptions, existing phylodynamic inference approaches assume that all observed and ancestral viral genetic variation is fitness-neutral. This assumption is violated more often than not, with a large body of analyses indicating that fitness varies substantially among genotypes circulating viral populations. Here, we focus specifically on fitness variation arising from deleterious mutations, asking whether incomplete purifying selection of deleterious mutations has the potential to bias phylodynamic inference. We use simulations of an exponentially growing population to explore how incomplete purifying selection distorts tree shape as well as how it shifts the distribution of non-neutral mutations over trees. Consistent with previous results, we find that incomplete purifying selection strongly shapes the distribution of mutations while only weakly impacting tree shape. Despite incomplete purifying selection shifting the distribution of mutations, we find little discernible bias in estimates of the viral growth rate and times of the most recent common ancestor. Our results reassuringly indicate that existing phylodynamic inference approaches may not yield biased epidemiological parameter estimates in the face of incomplete purifying selection, although more work is needed to assess the generalizability of these findings.