Advantages of a mechanistic codon substitution model for evolutionary analysis of protein-coding sequences

Area of Science:

• Evolutionary biology
• Molecular evolution
• Bioinformatics

Background:

• Existing models for analyzing protein-coding sequence evolution include nucleotide, amino acid, and empirical codon substitution models.
• A mechanistic codon substitution model, where substitution rates depend on mutation rates and amino acid fixation probabilities, offers advantages.
• This model can approximate diverse codon substitution processes and allows separate evaluation of nucleotide mutation and amino acid selection.

Purpose of the Study:

• To introduce and evaluate a novel mechanistic codon substitution model for evolutionary analysis.
• To compare the performance of the proposed model against existing nucleotide, amino acid, and empirical codon substitution models.
• To investigate the significance of multiple nucleotide changes and variation in selective constraints in phylogenetic analyses.

Main Methods:

• Developed a mechanistic codon substitution model incorporating codon mutation rates and amino acid replacement fixation probabilities.
• Allowed for multiple nucleotide changes within infinitesimal time, generalizing the time-reversible model.
• Tailored selective constraints to individual genes and estimated them by maximizing likelihoods of substitution frequency matrices.

Main Results:

• The proposed model significantly outperformed existing models across five phylogenetic trees using diverse gene sequences (chloroplast, mitochondrial, nuclear).
• Multiple nucleotide changes in infinitesimal time were found to be significant in long phylogenetic branches.
• Variation in selective constraint over sites provided a significantly better fit to datasets than variable mutation rates, except for slow-evolving mammalian nuclear genes.

Conclusions:

• The developed mechanistic codon substitution model provides a more accurate framework for evolutionary analysis of protein-coding sequences.
• The model's ability to separately assess mutation and selection is crucial for understanding evolutionary processes.
• Assuming variable mutation rates over sites can lead to inaccurate branch length estimations in phylogenetic analyses.