Adaptive distance measures for resolving K2P quartets: metric separation versus stochastic noise

Area of Science:

• Computational Biology
• Phylogenetics
• Bioinformatics

Background:

• Phylogenetic reconstruction relies on evolutionary distances derived from sequence data.
• Traditional methods assume a fixed evolutionary model, potentially limiting accuracy.
• Optimizing distance functions tailored to specific sequence sets is an active research area.

Purpose of the Study:

• To develop and present methods for selecting sequence-specific distance functions.
• To enhance the accuracy of phylogenetic trees reconstructed using distance-based methods.
• To improve the accuracy of quartet reconstruction within Kimura's 2-parameter model.

Main Methods:

• Investigating methods to adjust distance functions based on input DNA or protein sequences.
• Applying the four-point method for phylogenetic tree construction.
• Focusing on Kimura's 2-parameter model, including non-homogenous quartets.

Main Results:

• Demonstrated significant improvements in the accuracy of reconstructed phylogenetic quartets.
• Identified specific distance functions that maximize expected accuracy for given sequence sets.
• Showcased enhanced performance, particularly for non-homogenous evolutionary scenarios.

Conclusions:

• Tailoring distance functions to sequence data substantially boosts phylogenetic reconstruction accuracy.
• The proposed methods offer a more precise approach to inferring evolutionary histories.
• This work contributes to more reliable evolutionary analyses in computational biology.