Characterization of oligopeptide patterns in large protein sets

Area of Science:

• Bioinformatics
• Genomics
• Proteomics

Background:

• Oligopeptide patterns are crucial for understanding genome and kingdom-level classifications.
• Previous studies focused on habitat classification and local structural prediction using short peptide motifs.
• Characterizing informational peptide patterns is vital for new applications and avoiding bias pitfalls.

Purpose of the Study:

• To investigate four classes of pentapeptide patterns (POP, NEP, ORP, URP) across archaea, bacteria, and eukaryotes.
• To identify known and unknown sequence features within these pentapeptide pattern classes.
• To understand the inherent biases in naturally occurring oligopeptide patterns.

Main Methods:

• Analysis of four pentapeptide pattern classes: POP (statistically unexpected high abundance), NEP (statistically expected low abundance), ORP (kingdom-unique), and URP (kingdom-excluded).
• Utilized Swiss-Prot and a dataset of 386 completely sequenced genomes.
• Examined the 100 most extreme patterns within each class.

Main Results:

• Identified known and unknown sequence features within the studied pentapeptide patterns.
• Observed that most known motifs are explainable by their originating protein families.
• Found inherent biases in oligopeptide patterns not solely explained by residue distribution.

Conclusions:

• Identified three predominant categories of oligopeptide patterns: kingdom-widespread, structurally/functionally favored, and species-specific retrotransposons.
• These biases affect the accuracy of sequence pattern algorithms reliant on amino acid usage.
• The methods can identify kingdom-specific antigens for antibiotic targets and aid in detecting coding gene regions.