SAMP: Identifying Antimicrobial Peptides by an Ensemble Learning Model Based on Proportionalized Split Amino Acid...

Area of Science:

• Computational biology
• Biochemistry
• Infectious diseases

Background:

• Antimicrobial resistance (AMR) is a growing global health crisis, projected to cause 10 million deaths annually by 2050.
• Antimicrobial peptides (AMPs) are crucial components of the innate immune system with broad-spectrum activity against resistant pathogens.
• Current computational methods for AMP discovery often fail to fully capture complex sequence information.

Approach:

• Developed SAMP, an ensemble random projection (RP) model for predicting antimicrobial peptides (AMPs).
• Introduced Proportionalized Split Amino Acid Composition (PSAAC) as a novel feature set, complementing traditional sequence-based features.
• SAMP effectively captures residue patterns at peptide termini and sequence order information in internal fragments.

Key Points:

• SAMP demonstrates superior performance over existing methods like iAMPpred and AMPScanner V2 on various datasets.
• The model achieves higher accuracy, MCC, G-measure, and F1-score in AMP prediction.
• The ensemble RP architecture enhances scalability for large-scale AMP identification.

Conclusions:

• SAMP offers a significant advancement in computational AMP discovery, addressing limitations of previous approaches.
• The model's ability to leverage novel sequence features leads to more accurate predictions.
• A freely available Python package facilitates the adoption and application of SAMP in antimicrobial research.