Incorporating alignment uncertainty into Felsenstein's phylogenetic bootstrap to improve its reliability

Area of Science:

• Computational Biology
• Phylogenetics
• Bioinformatics

Background:

• Phylogenetic analyses often rely on pre-estimated multiple sequence alignments (MSAs).
• Previous work highlighted significant phylogenetic uncertainty introduced by different MSA methods.
• Distinguishing correct phylogenies from artifacts remains challenging.

Purpose of the Study:

• To develop a method that accounts for MSA-induced uncertainty in phylogenetic inference.
• To enhance the reliability of bootstrap values in phylogenetic analysis.
• To improve the accuracy of identifying correct phylogenetic clades.

Main Methods:

• Proposed Weighted Partial Super Bootstrap (wpSBOOT) method.
• Concatenated alternative MSAs from various aligners.
• Developed a bootstrap sampling strategy favoring unique alignment columns.

Main Results:

• wpSBOOT significantly increases the correlation between clade correctness and bootstrap values.
• Demonstrated improved predictive power on simulated datasets (16, 32, 64 tips).
• Achieved superior discrimination between correct and incorrect trees on empirical yeast data, with lower Type I error rates.

Conclusions:

• Incorporating MSA uncertainty via wpSBOOT enhances phylogenetic tree reliability.
• The method provides more accurate bootstrap support values.
• wpSBOOT offers a robust approach for phylogenetic reconstruction in the presence of alignment ambiguity.