BlaPred: Predicting and classifying β-lactamase using a 3-tier prediction system via Chou's general PseAAC

Area of Science:

• Biochemistry
• Computational Biology
• Drug Resistance

Background:

• Beta-lactam antibiotics are widely used, but beta-lactamase enzymes confer bacterial resistance.
• Existing methods for identifying and classifying beta-lactamases are time-consuming and resource-intensive.
• The rapid generation of genomic data necessitates efficient computational tools for enzyme analysis.

Purpose of the Study:

• To develop a fast and accurate computational system for predicting and classifying beta-lactamase enzymes from protein sequences.
• To provide a tool that aids in understanding bacterial resistance mechanisms in the post-genomic era.

Main Methods:

• Development of a three-tier prediction system, BlaPred, based on support vector machines.
• Utilized amino acid composition and pseudo amino acid compositions as input features.
• Employed leave-one-out cross-validation and independent datasets for performance evaluation.

Main Results:

• BlaPred achieved high accuracy in discriminating beta-lactamases from non-beta-lactamases (93.57% in tier-I).
• Accuracies for predicting Ambler classes A, B, C, and D were above 93%.
• Subclass prediction accuracies for B1, B2, and B3 were 84.78%, 95.65%, and 89.13%, respectively, with overall accuracy on independent datasets exceeding 92% for classes and 87% for subclasses.

Conclusions:

• BlaPred offers an efficient and accurate computational method for beta-lactamase identification and classification.
• The system significantly outperforms existing methods on benchmark datasets.
• BlaPred is available as a webserver and standalone software for broader accessibility.