GPCRsclass

GPCRsclass predicts and classifies amine-binding receptors of the amine subfamily of G-protein-coupled receptors (GPCRs) from protein sequences to support receptor identification and subtyping.

Key Features:

• Predictive methodology: Uses amino acid composition and dipeptide composition of protein sequences to predict amine-binding receptors, leveraging correlations between receptor types and composition as demonstrated by Chou and colleagues.
• Machine learning: Employs support vector machines (SVM) trained and tested on a dataset of 167 proteins from the amine subfamily of GPCRs.
• Performance metrics: Discrimination between the amine subfamily and globular proteins yields Matthew's correlation coefficients of 0.98 (amino acid composition) and 0.99 (dipeptide composition); classification accuracies for receptor types are 89.8% (amino acid composition) and 96.4% (dipeptide composition); and 67.6% of sequences are predicted with 100% accuracy for reliability index greater than or equal to 5.
• Validation: Model performance was evaluated using 5-fold cross-validation.

Scientific Applications:

• Identification of novel amine-type receptors: Enables detection of candidate amine-binding GPCR sequences for further study.
• Receptor subtype classification: Assigns amine GPCR sequences to specific receptor types to support subtype annotation.
• Drug discovery support: Provides computational predictions to inform early-stage receptor-targeted drug discovery.

Methodology:

The method analyzes amino acid and dipeptide compositions of protein sequences, applies support vector machines trained on known amine GPCR proteins, and uses 5-fold cross-validation for performance evaluation.