Characterization and identification of antimicrobial peptides with different functional activities

Area of Science:

• Biochemistry and Molecular Biology
• Computational Biology and Bioinformatics
• Drug Discovery and Development

Background:

• Antimicrobial peptides (AMPs) are vital components of the innate immune system with broad-spectrum activity against pathogens.
• Existing machine learning methods for AMP prediction often overlook the diverse functional activities of these peptides.
• Understanding sequence features associated with specific AMP activities is crucial for therapeutic development.

Purpose of the Study:

• To develop a systematic scheme for identifying antimicrobial peptides (AMPs) and characterizing their diverse functional activities.
• To investigate sequence-based features that differentiate AMPs with various activities, including anti-parasitic, anti-viral, anti-cancer, anti-fungal, and activity against different bacterial types and mammals.
• To create a user-friendly web server for exploring AMP functions.

Main Methods:

• A two-stage machine learning approach was proposed: the first stage identifies AMPs, and the second characterizes their functional activities.
• Sequential forward selection was used to extract informative sequence features such as hydrophobicity, normalized van der Waals volume, polarity, charge, and solvent accessibility.
• Pseudo amino acid composition was identified as a key feature for differentiating AMP functional activities.

Main Results:

• The first-stage AMP classifier achieved a high Area Under the Receiver Operating Characteristic Curve (AUC) of 0.9894.
• The second-stage multi-class models demonstrated strong performance in classifying AMPs by functional activity, with AUCs ranging from 0.7773 to 0.9404.
• The study successfully implemented a web server (AMPfun) to facilitate the exploration of AMP functions.

Conclusions:

• The proposed two-stage scheme effectively identifies antimicrobial peptides (AMPs) and accurately characterizes their diverse functional activities.
• Sequence-derived features, including pseudo amino acid composition, are critical for understanding and predicting AMP functionality.
• The AMPfun web server provides a valuable resource for researchers investigating antimicrobial peptide therapeutics.