AntiTbPred

AntiTbPred predicts peptides with bactericidal activity against Mycobacterium species to identify antitubercular peptides for tuberculosis research.

Key Features:

• Sequence analysis and feature extraction: Analyzes peptide sequences and identifies residue patterns, including Lysine, Arginine, Leucine, and Tryptophan, associated with antitubercular activity.
• Machine learning models: Implements support vector machine (SVM) models trained on amino acid composition, binary profile of terminus residues (N5C5), and dipeptide composition.
• Ensemble classifiers: Combines amino acid composition and N5C5 binary-pattern models, achieving Acc 73.20% and AUROC 0.80 on the main dataset and Acc 75.62% and AUROC 0.83 on a secondary dataset.
• Hybrid model: Uses a hybrid model that attains Acc 75.87% and AUROC 0.83 on the main dataset and Acc 78.54% and AUROC 0.86 on the secondary dataset.

Scientific Applications:

• Peptide design: Facilitates prediction and design of novel antitubercular peptides.
• Tuberculosis research and drug development: Supports identification of peptide candidates targeting Mycobacterium species for therapeutic investigation.
• Antibiotic resistance research: Aids exploration of peptide-based alternatives to traditional antibiotics for drug-resistant TB.

Methodology:

Sequence-derived features—amino acid composition, dipeptide composition, and N5C5 binary profiles of N- and C-terminus residues—were used as inputs to support vector machine (SVM) models, ensemble classifiers, and a hybrid model.