AMPFinder: A computational model to identify antimicrobial peptides and their functions based on sequence-derived...

Area of Science:

• Biochemistry
• Computational Biology
• Drug Discovery

Background:

• Antimicrobial peptides (AMPs), or host defense peptides, are vital components of innate immunity across all life forms.
• AMPs exhibit broad-spectrum activity against bacteria, viruses, fungi, and cancer cells, and importantly, are not prone to drug resistance.
• The urgent need for high-throughput methods to identify AMPs and predict their functions is driven by their therapeutic potential.

Purpose of the Study:

• To develop and validate a novel computational model, AMPFinder, for the accurate identification and functional classification of antimicrobial peptides.
• To leverage sequence-derived features and language embedding for enhanced AMP prediction capabilities.

Main Methods:

• A cascaded computational model, AMPFinder, was designed using sequence-derived and language embedding features.
• The model's performance was evaluated against state-of-the-art methods on independent test and public datasets.
• Statistical metrics including F1-score, Matthews Correlation Coefficient (MCC), Area Under the Curve (AUC), R-squared bias, and Average Precision (AP) were used for comparison.

Main Results:

• AMPFinder demonstrated superior performance in both AMP identification and function prediction compared to existing methods.
• Significant improvements were observed in F1-score (1.45%-6.13%), MCC (2.92%-12.86%), AUC (5.13%-8.56%), and AP (9.20%-21.07%) on an independent test dataset.
• AMPFinder achieved a notable reduction in R-squared bias (18.82%-19.46%) on a public dataset via 10-fold cross-validation, indicating higher accuracy and reliability.

Conclusions:

• AMPFinder provides a highly accurate and efficient computational approach for identifying antimicrobial peptides and predicting their functional types.
• The model's superior performance suggests its utility as a valuable tool for accelerating the discovery of novel AMP-based therapeutics.
• The study makes datasets, source code, and a user-friendly application publicly available to facilitate further research and application.