PTPAMP: prediction tool for plant-derived antimicrobial peptides

Area of Science:

• Computational Biology
• Biochemistry
• Plant Science

Background:

• Antimicrobial peptides (AMPs) are crucial in innate immunity and represent alternatives to conventional antibiotics.
• Plants possess numerous AMP-encoding genes, necessitating specialized prediction tools for plant-derived AMPs.

Purpose of the Study:

• To develop a computational tool for predicting plant-derived antimicrobial peptides (AMPs).
• To classify predicted peptides into four functional activities: antimicrobial (AMP), antibacterial (ABP), antifungal (AFP), and antiviral (AVP).

Main Methods:

• Developed predictive models using peptide features like amino acid composition, dipeptide composition, and physicochemical attributes.
• Integrated models into a user-friendly web server named PTPAMP.
• Evaluated model performance on benchmark datasets.

Main Results:

• Achieved high prediction accuracy with average AUCs of 0.95 (AMP), 0.91 (ABP), 0.85 (AFP), and 0.88 (AVP).
• Performance exceeded state-of-the-art methods by approximately 6.75%.
• Identified distinct residue patterns, including cysteine abundance, in plant-derived AMPs.

Conclusions:

• PTPAMP is an effective tool for high-throughput identification of plant-derived AMPs.
• The findings provide insights into the structural characteristics of plant AMPs.
• The web server is publicly accessible for research use.