iBitter-Fuse

iBitter-Fuse predicts bitter peptides from peptide and protein sequences to support identification and de novo design of bitter peptides for applications in pharmacology, nutrition, and biochemistry.

Key Features:

• Multi-View Feature Integration: Integrates various feature encoding schemes that capture compositional information and physicochemical properties of peptides.
• Feature Encoding Schemes: Applies comprehensive encoding to represent a wide range of peptide characteristics.
• Customized Genetic Algorithm (GA-SAR): Uses a genetic algorithm with self-assessment-reporting (GA-SAR) to identify and select informative features.
• Support Vector Machine (SVM) Classifier: Employs an SVM-based classifier to construct the final predictive model from selected features.
• Performance Evaluation: Evaluates predictive performance using 10-fold cross-validation and independent tests.

Scientific Applications:

• High-throughput identification: Enables large-scale screening and identification of bitter peptides from protein datasets.
• Discovery and de novo design: Supports discovery and rational design of novel bitter peptides.
• Domain-specific studies: Facilitates research applications in pharmacology, nutrition, and biochemistry by providing predictive annotations of bitterness.

Methodology:

Feature encoding to capture compositional and physicochemical properties; feature selection via a genetic algorithm with self-assessment-reporting (GA-SAR); model development using a support vector machine (SVM); performance assessed by 10-fold cross-validation and independent tests.