deep-AMPpred: A Deep Learning Method for Identifying Antimicrobial Peptides and Their Functional Activities.

Area of Science:

• Bioinformatics
• Computational Biology
• Drug Discovery

Background:

• Antimicrobial peptides (AMPs) are vital for innate immunity and are increasingly important as alternatives to conventional antibiotics due to rising pathogen resistance.
• Current computational methods for AMP identification often lack a focus on predicting diverse functional activities, limiting the discovery of broad-spectrum candidates.

Purpose of the Study:

• To develop a novel computational tool, deep-AMPpred, for accurate identification of antimicrobial peptides (AMPs).
• To predict multiple functional activities of AMPs, facilitating the discovery of peptides with broad-spectrum antimicrobial capabilities.

Main Methods:

• A two-stage prediction framework was employed: the first stage distinguishes AMPs from non-AMPs, and the second stage performs multilabel classification for 13 common AMP functional activities.
• The deep-AMPpred model integrates the ESM-2 language model for peptide sequence feature encoding with Convolutional Neural Networks (CNN), Bidirectional Long Short-Term Memory (BiLSTM), and Convolutional Block Attention Module (CBAM) for enhanced prediction.

Main Results:

• The deep-AMPpred model demonstrated high performance in accurately identifying AMPs.
• The tool successfully predicted multiple functional activities of AMPs, highlighting its capability in functional characterization.

Conclusions:

• The study validates the effectiveness of the ESM-2 model in capturing essential peptide sequence features for AMP prediction.
• Integrating multiple deep learning architectures (CNN, BiLSTM, CBAM) significantly improves the performance of AMP identification and functional activity prediction.