Statistical phylogenetic tree analysis using differences of means.

Area of Science:

• Phylogenetics
• Computational Biology
• Evolutionary Biology

Background:

• Phylogenetic trees are crucial for understanding evolutionary relationships.
• Assessing congruence between trees derived from different datasets is vital for accurate evolutionary inference.
• Existing methods often rely on point estimates, potentially overlooking nuances in tree distributions.

Purpose of the Study:

• To introduce a novel statistical method for testing significant incongruence between two phylogenetic trees based on their alignments.
• To provide a robust framework for analyzing gene tree incongruence and detecting evolutionary anomalies.

Main Methods:

• The method compares distributions of phylogenetic trees derived from two alignments, rather than point estimates.
• Trees are mapped into a vector space, and the difference of means is used for comparison.
• Kernel methods accelerate distance calculations in high-dimensional feature spaces (e.g., splits, quartets).
• Bootstrapping of alignment columns is employed to determine statistical significance (p-values).

Main Results:

• The statistical method was validated using simulated data under a coalescence model.
• The approach successfully identified incongruence in simulated datasets designed to test gene tree congruence.
• Applications to real-world datasets (gophers/lice, grasses/endophytes, fungal genes) demonstrated practical utility.

Conclusions:

• The proposed statistical method offers a powerful tool for assessing phylogenetic tree incongruence.
• This approach enhances the detection of complex evolutionary events, such as horizontal gene transfer and genome reshuffling.
• The accompanying Phylotree toolkit facilitates the application and further development of this method.