Tip dating of phylogenetic trees is a growing discipline and represents a powerful tool for hypothesis testing. It uses sequence data isolated at different points in time (i.e., heterochronous datasets) to co-estimate the timing of evolutionary and ecological events with rates of molecular evolution. For example, such analytical approaches have been critical in dating key events in human evolutionary history or improving our understanding of the context of the emergence of various important pathogens.

Inferences from tip-calibrated phylogenetic trees are only valid if there is enough and consistent temporal signal in the data. As such, it is very important to test the molecular data for the amount of temporal signal and its consistency among various samples prior to any tip-dating inference. Two tests aiming to evaluate the reliability of Bayesian inferences from time-structured data have been previously developed: the “date-randomization test” (DRT hereafter) and the “leave-one-out cross-validation” (LOOCV hereafter). These two tests are critical to determine the presence of temporal signal and correctness of tip dates contained in the datasets. TipDatingBeast is an R package built to assist performing such preliminary analysis for tip-dating using the BEAST program. In this workshop we will provide a general introduction of current approaches, demonstrate how to use the functions with an empirical dataset, and discuss and guide the results interpretation. Participants will have the opportunity to learn a new set of tools and discuss potential new applications for their own research interests.