Summarizing a posterior distribution of trees using agreement subtrees.

Area of Science:

• Computational Biology
• Evolutionary Biology
• Bioinformatics

Background:

• Bayesian phylogenetic inference generates a posterior distribution of trees, offering more information than a single best tree.
• Current methods often summarize this distribution with a majority-rule consensus tree, which can discard valuable data.
• This approach can oversimplify complex phylogenetic signals.

Purpose of the Study:

• To develop a novel method for summarizing Bayesian posterior distributions of phylogenetic trees.
• To present an alternative to the majority-rule consensus tree that better represents the full distribution.
• To provide a more informative summary for phylogenetic analysis.

Main Methods:

• Identifying agreement subtrees that are consistently found within the posterior distribution of trees.
• Developing algorithms to efficiently search and extract these frequently occurring subtrees.
• Annotating these subtrees with measures of their prevalence in the posterior distribution.

Main Results:

• The proposed method generates fully resolved binary trees for subsets of taxa.
• These trees are associated with high marginal posterior probabilities.
• The method captures information about the spread and uncertainty within the posterior distribution.

Conclusions:

• Summarizing Bayesian phylogenetic trees using agreement subtrees offers a more comprehensive view than traditional methods.
• This approach preserves more information from the posterior distribution, leading to potentially more robust evolutionary inferences.
• The method provides a valuable tool for exploring phylogenetic uncertainty and identifying well-supported clades.