Pred-BVP-Unb

Pred-BVP-Unb predicts bacteriophage virion proteins (BVPs) from protein sequences to enable identification of virion components relevant to genetic engineering and phage therapy.

Key Features:

• Bi-PSSM Evolutionary Information: Captures evolutionary patterns and relationships among proteins using Bi-PSSM-derived information.
• Composition & Translation: Analyzes protein composition and translation-related properties.
• Split Amino Acid Composition: Provides detailed insights into amino acid distribution across protein segments.
• Addressing Class Imbalance: Manages class imbalance using a synthetic minority oversampling technique.
• Feature Selection via Recursive Feature Elimination (RFE): Employs RFE to identify and select essential attributes and optimize the feature space.
• Support Vector Machine Classifier: Uses a support vector machine classifier with a radial base kernel for prediction.

Scientific Applications:

• Accelerated Identification: Accelerates identification of BVPs within large protein datasets.
• Enhanced Accuracy: Reports 92.54% accuracy on benchmark datasets and 83.06% on independent datasets.
• Drug Design and Discovery: Provides BVP predictions that inform antibacterial drug design and discovery of new bacteriophage virion proteins.

Methodology:

Feature representation uses Bi-PSSM evolutionary information, Composition & Translation, and Split Amino Acid Composition; class imbalance is handled via a synthetic minority oversampling technique; features are selected with recursive feature elimination (RFE); classification is performed by a support vector machine with a radial base kernel.