antibp2

antibp2 predicts antibacterial peptides within protein sequences using Support Vector Machine (SVM) models that assess amino acid composition and binary patterns at peptide termini to identify and classify antibacterial activity.

Key Features:

• Terminal-pattern model: Uses binary patterns of the first 15 residues at the N-terminus and C-terminus and combined termini to predict antibacterial activity with 91.64% accuracy and an MCC of 0.831.
• Composition-based SVM: Employs an SVM model based on whole amino acid composition achieving 92.14% accuracy and an MCC of 0.843.
• Validation: Models were validated using five-fold cross-validation and performance was assessed on independent datasets.
• Classification capabilities: Classifies peptides according to their sources with 98.95% accuracy and assigns peptides to specific families.
• Training dataset: Developed from analysis of 999 antibacterial peptides sourced from the Antibacterial Peptide Database (APD).

Scientific Applications:

• Drug Discovery: Identification of peptide candidates with antibacterial properties to support development of novel therapeutics against bacterial infections.
• Immunology Research: Analysis of residue contributions to antibacterial activity to inform studies of innate immune effectors.
• Peptide Engineering: Guiding design and modification of peptides for enhanced or specific antibacterial functions based on predicted features.

Methodology:

Analysis of 999 antibacterial peptides from the Antibacterial Peptide Database (APD); identification of preferred residues at N- and C-termini using binary patterns of the first 15 residues; development of SVM models using binary patterns and amino acid composition; validation via five-fold cross-validation and testing on independent datasets.