Robustness of maximum likelihood tree estimation against different patterns of base substitutions

Area of Science:

• Molecular Evolution
• Phylogenetics
• Bioinformatics

Background:

• The Maximum Likelihood (ML) method is widely used for molecular phylogenetic tree estimation.
• Simplifying assumptions about nucleotide substitution patterns are often made due to the unknown complexity of evolutionary processes.
• The robustness of these simplifications in ML tree estimation has not been fully guaranteed.

Purpose of the Study:

• To evaluate the robustness of the ML method in molecular tree estimation against various nucleotide substitution patterns.
• To assess the impact of different transition/transversion ratios and GC content on ML tree accuracy.
• To compare the performance of the simplest Jukes-Cantor model with more complex models in phylogenetic inference.

Main Methods:

• Computer simulations were conducted using hypothetical genes with defined evolutionary models and true trees.
• The ML method was applied using different nucleotide substitution models, including Jukes-Cantor.
• The accuracy of estimated tree topology and branch lengths was compared to the true trees.
• The ML method was applied to real nucleotide sequence data from T-cell leukemia virus tax genes.

Main Results:

• ML topology estimation demonstrated considerable robustness against variations in transition/transversion ratios and GC content.
• Branch length estimation in ML trees was found to be less robust to these variations.
• ML tree estimation using the Jukes-Cantor model showed resistance to GC content but sensitivity to transition/transversion ratios.
• Simulation results were consistent with analyses of T-cell leukemia virus data, indicating the Jukes-Cantor model's utility.

Conclusions:

• The ML method's topology estimation is reliable across different nucleotide substitution patterns.
• Branch length estimation requires careful consideration of the chosen substitution model.
• The Jukes-Cantor model is a practical and useful tool for inferring molecular phylogeny, even for complex evolutionary histories.