A58 Epidemic dynamics of ancient disease outbreaks

Bayesian phylogenetic analysis allows for the estimation of the time to the most recent common ancestor (tMRCA) of sequences sampled at different times, as long as they prove to be ‘measurably evolving’, which means that the time between sampling dates was long enough to allow the appearance of a measurable amount of genetic changes. This ‘temporal signal’ can be tested with the software TempEst (Rambaut et al. 2016), which generates a regression of the root-to-tip genetic distance on sampling times and finds the best-fitting root that produces the lowest residual sum of squares. For the case of pathogen single nucleotide polymorphism (SNP) alignments, containing both modern and ancient sequences, it is common to find positions with unknown nucleotides (gaps) that could generate problems in the phylogenetic reconstruction. Thus, the use of complete deletion alignments is fairly common. This practice, however, could cause the loss of potentially important information, so we aim to identify the most suitable deletion threshold for the proportion of unknown sites allowed for a given alignment before proceeding to analyze the data in BEAST. Here, I present the temporal signal of 204 whole-genome sequences of Yersinia pestis, a zoonotic gram-negative bacteria and causal agent of the bubonic, pneumonic, and systemic plagues. I demonstrate measurable temporal signal for the alignment with thresholds of 0–10 per cent for the proportion of unknown sites per SNP. The results showed that a complete deletion alignment presented the lowest correlation and greatest residual mean squared values. The best threshold depends on the method used to find the best root, but appears to be between 7–9 per cent.