Detecting evolutionary trends from molecular data. 1. Some measures of compositional nonrandomness.

Area of Science:

• Biochemistry
• Evolutionary Biology
• Bioinformatics

Background:

• Protein composition analysis is crucial for understanding evolutionary processes.
• Quantifying nonrandomness in amino acid sequences aids in identifying functional constraints.
• Previous methods may lack the resolution to detect subtle evolutionary changes.

Purpose of the Study:

• To assess the physical significance and evolutionary detection power of compositional nonrandomness measures.
• To compare different levels of analysis, including optimal, integer, noise, and real protein compositions.
• To identify the most effective measure for analyzing protein sequence variations.

Main Methods:

• Defined four levels for compositional analysis: base (optimal non-integer), integer (optimal integer), noise (random), and real protein.
• Utilized a priori probabilities (pi) and amino acid occurrences (ni) for calculating nonrandomness.
• Evaluated the measure S for its physical sense and resolution across different levels.

Main Results:

• The measure S exhibits the most direct physical significance.
• S demonstrates the smoothest behavior and smallest relative fluctuations across all defined levels.
• S provides the highest resolution for detecting evolutionary significant variations in protein composition.

Conclusions:

• The measure S is superior for analyzing protein compositional nonrandomness.
• S offers enhanced power in detecting evolutionary variations compared to other measures.
• This approach provides a robust framework for studying protein evolution through compositional analysis.