AMPpred-MFA: An Interpretable Antimicrobial Peptide Predictor with a Stacking Architecture, Multiple Features, and...

Area of Science:

• Biochemistry and Molecular Biology
• Computational Biology and Bioinformatics
• Infectious Diseases

Background:

• Antimicrobial peptides (AMPs) are crucial for combating microbial infections.
• Existing computational models for AMP identification lack high performance and interpretability.
• There is a need for advanced computational tools to discover and predict AMPs.

Purpose of the Study:

• To develop a high-performance and interpretable computational model for identifying antimicrobial peptides (AMPs).
• To leverage deep learning architectures and advanced feature representations for improved AMP prediction.

Main Methods:

• Collected and processed new benchmark datasets for AMP identification.
• Designed a stacking deep architecture, AMPpred-MFA, integrating bidirectional LSTM and CNN.
• Utilized multiple sequence features and a multihead attention mechanism within the model.

Main Results:

• AMPpred-MFA demonstrated state-of-the-art performance in identifying AMPs across five independent tests.
• Visualization and ablation studies confirmed the model's effectiveness and the importance of its components.
• The model's feature representation and stacking architecture, particularly the multihead attention, were validated.

Conclusions:

• AMPpred-MFA provides a reliable and efficient approach for understanding and predicting antimicrobial peptides.
• The developed model addresses the limitations of existing computational methods for AMP discovery.
• This work contributes to the advancement of computational strategies in antimicrobial research.