SAMP: Identifying antimicrobial peptides by an ensemble learning model based on proportionalized split amino acid...

Area of Science:

• Computational biology
• Drug discovery
• Bioinformatics

Background:

• Drug-resistant bacterial infections pose a significant global health threat, projected to cause 10 million deaths by 2050.
• Antimicrobial peptides (AMPs) are crucial innate immune effectors with broad-spectrum activity against resistant pathogens.
• Current computational methods for AMP discovery often overlook critical sequence information.

Purpose of the Study:

• To develop an advanced computational model for accurate prediction of antimicrobial peptides (AMPs).
• To introduce novel sequence-based features that capture comprehensive peptide information.
• To enhance the scalability and performance of AMP identification tools.

Main Methods:

• Developed SAMP, an ensemble random projection (RP) based model.
• Incorporated proportionalized split amino acid composition (PSAAC) alongside conventional sequence features.
• Utilized N-terminal, C-terminal, and middle fragment sequence order information.

Main Results:

• SAMP demonstrated superior performance over existing methods like iAMPpred and AMPScanner V2.
• Achieved higher accuracy, Matthews correlation coefficient (MCC), G-measure, and F1-score on benchmark datasets.
• The ensemble RP architecture ensures scalability for large-scale AMP identification.

Conclusions:

• SAMP offers a robust and accurate computational approach for identifying novel antimicrobial peptides.
• The novel PSAAC feature set effectively captures critical sequence patterns for improved prediction.
• SAMP provides a scalable and high-performing solution to accelerate the discovery of new antibiotics.