Estimating the Effective Sample Size of Tree Topologies from Bayesian Phylogenetic Analyses

Area of Science:

• Computational Biology
• Evolutionary Biology
• Statistical Modeling

Background:

• Bayesian phylogenetic analyses rely on Markov chain Monte Carlo (MCMC) methods to estimate posterior distributions of tree topologies and parameters.
• Assessing the adequacy of these posterior distributions is crucial for reliable inferences.
• Effective sample size (ESS) quantifies independent samples but lacks a method for tree topologies, a primary parameter of interest.

Purpose of the Study:

• To develop and present methods for estimating the ESS of phylogenetic tree topologies.
• To introduce new diagnostic plots for evaluating posterior samples of tree topologies.
• To provide tools for assessing the mixing and convergence of tree topologies in Bayesian MCMC analyses.

Main Methods:

• Development of novel methods to calculate ESS specifically for tree topology samples.
• Integration of these ESS estimation methods with two new diagnostic visualization tools.
• Performance evaluation using both simulated and empirical phylogenetic datasets.

Main Results:

• Successful development of methods to estimate ESS for tree topologies.
• Introduction of diagnostic plots that aid in assessing tree topology sampling adequacy.
• Demonstration of the utility of the new methods on diverse datasets.

Conclusions:

• The presented methods offer a solution to the critical need for assessing tree topology sampling in Bayesian phylogenetics.
• These tools enhance the reliability of phylogenetic inferences by ensuring adequate exploration of the tree topology space.
• The new diagnostics improve the assessment of mixing and convergence for tree topologies in MCMC analyses.