ACEP

ACEP classifies antibacterial peptides using deep learning to learn amino acid embedding patterns and attention scores for high-throughput antimicrobial peptide (AMP) recognition and motif discovery.

Key Features:

• Deep Learning Architecture: ACEP employs multi-layer neural networks that learn amino acid embedding patterns to automatically extract sequence features.
• Heterogeneous data fusion: The model fuses peptide sequence information and Position-Specific Scoring Matrix (PSSM) profiles to integrate diverse feature types.
• High-throughput recognition: ACEP processes large peptide datasets for extensive screening and classification of AMPs.
• Motif discovery and visualization: The method identifies important sequence motifs and provides visualization of data patterns and attention scores across model layers.
• Attention mechanism: ACEP calculates attention scores to highlight sequence regions that contribute most to its predictions.

Scientific Applications:

• Novel AMP discovery: ACEP supports identification of candidate antimicrobial peptides for downstream experimental validation.
• Resistance mitigation research: The classifier facilitates exploration of AMPs as alternatives to conventional antibiotics to address antimicrobial resistance.
• Bioinformatics studies: ACEP enables integrated analyses of sequence and PSSM data to study peptide function, motifs, and classification performance.

Methodology:

ACEP implements a deep learning framework with multiple neural network layers, uses amino acid embeddings, integrates sequence and PSSM-derived features, and applies an attention mechanism to extract sequence features and compute attention scores for motif identification.