The Bayesian Phylogenetic Bootstrap and its Application to Short Trees and Branches

Area of Science:

• Phylogenetics
• Computational Biology
• Evolutionary Biology

Background:

• Felsenstein's bootstrap is standard for phylogenetic branch support but struggles with massive, similar sequence data (e.g., SARS-CoV-2).
• Frequentist bootstrap sampling yields lower expected support (∼63%) for single-mutation branches, despite their high likelihood.
• This limitation hinders accurate phylogenetic inference with modern high-throughput sequencing.

Purpose of the Study:

• To develop a Bayesian phylogenetic bootstrap method addressing limitations of the frequentist approach.
• To provide a more accurate and interpretable measure of branch support for large-scale phylogenetics.
• To re-evaluate phylogenetic branch support by treating sequence alignments as complete information sources.

Main Methods:

• Proposed a Bayesian phylogenetic bootstrap by assigning uninformative prior probabilities to sites.
• Derived formulas for expected branch supports under perfect phylogeny in both frequentist and Bayesian frameworks.
• Validated theoretical results through simulations and analyses on viral and nonviral datasets.

Main Results:

• The Bayesian bootstrap assigns branch support as a posterior probability, offering clearer interpretation.
• Expected support for a single-mutation branch increases to ∼90% in the Bayesian framework.
• Simulations confirm the robustness of theoretical findings, with Bayesian support being more interpretable for low-homoplasy data.

Conclusions:

• The proposed Bayesian phylogenetic bootstrap provides more reliable and interpretable branch support, particularly for large, similar sequence datasets.
• This method enhances phylogenetic accuracy for rapidly evolving organisms like viruses.
• Bayesian bootstrap support aligns better with the high confidence expected for correct evolutionary inferences.