Speaker
Description
When constructing phylogenetic trees for phylodynamic and phylogeographic analyses, researchers often rely on tree-building software such as IQ-TREE, BEAST, and BEAST2. The algorithms used by these tools, whether based on maximum likelihood or Bayesian principles, are inherently stochastic. As a result, repeated analyses of the same data can yield trees with unstable structures due to weak or noisy phylogenetic signals. This instability can significantly impact downstream inferences by obscuring uncertainties introduced by algorithmic stochasticity.
We aim to raise awareness about the importance of repeating the tree-building process and evaluating the stability of results. Using empirical examples based on published HIV data, we discuss approaches ranging from quick visual checks to more robust stability analyses, highlighting common pitfalls. Finally, we touch on work-in-progress methods and the challenges currently faced in their development.
| Expedited Notification | No thanks, I do not require Expedited Notification |
|---|
