A simple method for data partitioning based on relative evolutionary rates.

Area of Science:

• Phylogenetics and Evolutionary Biology
• Computational Biology and Bioinformatics

Background:

• Model-based phylogenetic analyses benefit from effective data partitioning.
• Traditional partitioning relies on a priori assumptions about sequence evolution, such as codon positions.
• Existing methods may not optimally handle evolutionary rate heterogeneity within datasets.

Purpose of the Study:

• To introduce a novel data partitioning method based on relative evolutionary rates of sites.
• To assess the performance of this new method using simulations and real-world phylogenetic datasets.
• To compare the new method against established partitioning strategies.

Main Methods:

• Relative evolutionary rates were inferred using Tree Independent Generation of Evolutionary Rates (TIGER).
• A novel Python script, RatePartitions, was developed to partition data based on these inferred rates.
• The method was evaluated using simulations and applied to eight Lepidoptera multi-locus datasets and one phylogenomic dataset.

Main Results:

• Simulations showed no adverse effects of the partitioning approach, even with missing data or challenging parameters.
• Partitioning with TIGER-rates and RatePartitions significantly improved phylogenetic analyses, as indicated by BIC scores, across all eight Lepidoptera datasets.
• The method demonstrated superior performance compared to standard partitioning strategies like gene- or codon-based approaches.

Conclusions:

• A new, data-driven partitioning method has been developed that does not require prior knowledge of sequence evolution.
• The method is versatile and applicable to various data types, including DNA, protein, and morphological characters.
• Improved performance is attributed to a greater account of data heterogeneity compared to prior knowledge-based methods.