AtbPpred: A Robust Sequence-Based Prediction of Anti-Tubercular Peptides Using Extremely Randomized Trees.

Area of Science:

• Biochemistry
• Computational Biology
• Drug Discovery

Background:

• Tuberculosis (TB), caused by Mycobacterium tuberculosis, affects a third of the global population.
• Rising multidrug-resistant and extensively drug-resistant TB necessitates novel therapeutic strategies.
• Peptide-based therapies offer advantages like targeted action and low immunogenicity.

Purpose of the Study:

• To develop an efficient computational method for predicting anti-tubercular peptides (AtbPs).
• To overcome the limitations of laborious and expensive experimental identification of AtbPs.
• To accelerate the discovery of new anti-TB peptide drugs.

Main Methods:

• Developed a two-layer machine learning (ML) predictor, AtbPpred.
• Employed a two-step feature selection and extremely randomized tree (ERT) models in the first layer.
• Integrated predictions from nine models into a final ERT model for enhanced accuracy.

Main Results:

• AtbPpred achieved 88.3% accuracy during cross-validation and 87.3% in independent evaluation.
• The predictor outperformed the state-of-the-art method by approximately 8.7% and 10.0%.
• A user-friendly webserver for AtbPpred is available at http://thegleelab.org/AtbPpred.

Conclusions:

• AtbPpred offers a high-throughput solution for identifying potential anti-tubercular peptides.
• The tool can aid in mechanistic insights into peptide functions against Mycobacterium tuberculosis.
• This computational approach is crucial for developing new therapies against resistant TB strains.