Empirical substitution models of protein evolution: database, relationships, and modeling considerations

Area of Science:

• Computational Biology
• Evolutionary Biology
• Bioinformatics

Background:

• Substitution models are crucial for protein evolution studies and phylogenetic inference.
• Empirical models are widely used in phylogenetics due to their simplicity, but a centralized database is lacking.
• Formal comparisons between existing empirical models are scarce, hindering assessment and implementation.

Purpose of the Study:

• To introduce EModelDB, a comprehensive database of empirical substitution models for protein evolution.
• To provide exchangeability matrices, model classification, and biological information for each model.
• To facilitate the assessment and implementation of these models in phylogenetic analyses.

Main Methods:

• Developed EModelDB with a user-friendly graphical interface using Python and SQL.
• Compiled a database of empirical substitution models, including exchangeability matrices and classifications.
• Compared common empirical models based on substitution rates and equilibrium amino acid frequencies.

Main Results:

• Substitution models derived from evolutionarily related proteins cluster together, indicating similar evolutionary patterns.
• Empirical models generally predict less stable proteins compared to real proteins.
• EModelDB provides a centralized resource for accessing and evaluating protein substitution models.

Conclusions:

• EModelDB enhances accessibility and assessment of empirical protein substitution models for phylogenetics.
• Understanding model-specific evolutionary patterns and their impact on protein stability is important.
• Models incorporating additional evolutionary constraints may be preferable for studying protein evolution with folding stability considerations.