nT4X and nT4M: Novel Time Non-reversible Mixture Amino Acid Substitution Models

Area of Science:

• Molecular Evolution
• Computational Biology
• Bioinformatics

Background:

• Modeling amino acid substitutions is crucial but challenging in molecular evolution research.
• Single-matrix models like LG are popular but may not capture site-specific substitution heterogeneity.
• Existing mixture models often assume evolutionary process reversibility for computational ease, potentially limiting biological realism.

Purpose of the Study:

• To develop more biologically realistic models for amino acid substitutions by relaxing the reversibility assumption.
• To introduce novel non-reversible mixture models for improved evolutionary inferences.
• To enable the inference of rooted phylogenetic trees using more accurate evolutionary models.

Main Methods:

• Developed two general non-reversible 4-matrix mixture models: nT4M and nT4X.
• Estimated model parameters using amino acid alignments from HSSP and TreeBASE databases.
• Compared the performance of the new non-reversible models against single-matrix and reversible mixture models.

Main Results:

• The newly estimated non-reversible models (nT4M, nT4X) demonstrated superior performance compared to single-matrix models.
• These non-reversible models outperformed most existing reversible mixture models in accuracy.
• The developed models facilitate the inference of rooted phylogenetic trees, a capability lacking in many current models.

Conclusions:

• Relaxing the assumption of reversibility leads to more biologically realistic and accurate models of amino acid substitution.
• The nT4M and nT4X models represent a significant advancement in molecular evolution modeling.
• Non-reversible mixture models enhance phylogenetic inference, particularly for rooted trees.