Estimating evolutionary rates using time-structured data: a general comparison of phylogenetic methods

Bioinformatics. 2016 Nov 15;32(22):3375-3379. doi: 10.1093/bioinformatics/btw421. Epub 2016 Jul 13.


Motivation: In rapidly evolving pathogens, including viruses and some bacteria, genetic change can accumulate over short time-frames. Accordingly, their sampling times can be used to calibrate molecular clocks, allowing estimation of evolutionary rates. Methods for estimating rates from time-structured data vary in how they treat phylogenetic uncertainty and rate variation among lineages. We compiled 81 virus data sets and estimated nucleotide substitution rates using root-to-tip regression, least-squares dating and Bayesian inference.

Results: Although estimates from these three methods were often congruent, this largely relied on the choice of clock model. In particular, relaxed-clock models tended to produce higher rate estimates than methods that assume constant rates. Discrepancies in rate estimates were also associated with high among-lineage rate variation, and phylogenetic and temporal clustering. These results provide insights into the factors that affect the reliability of rate estimates from time-structured sequence data, emphasizing the importance of clock-model testing.

Contact: or garzonsebastian@hotmail.comSupplementary information: Supplementary data are available at Bioinformatics online.

MeSH terms

  • Bacteria* / genetics
  • Bayes Theorem
  • Models, Genetic
  • Phylogeny*
  • Reproducibility of Results
  • Viruses* / genetics