GRIFFIN

GRIFFIN predicts interactions between G-protein coupled receptors (GPCRs) and G-proteins and determines G-protein coupling selectivity for biological and pharmacological analysis.

Key Features:

• Predictive Accuracy: Reports average accuracy exceeding 85% with high sensitivity and specificity for predicted GPCR–G-protein pairings.
• Algorithms: Uses support vector machine (SVM) algorithms and hidden Markov models (HMMs) for prediction.
• Structural Feature Vectors: Derives quantitative features from entire structural segments of ligands, GPCRs, and G-proteins to construct feature vectors for classifiers.
• Hierarchical SVM Classifier: Employs a hierarchical SVM classifier specifically for Class A GPCRs.
• HMM Utilization: Applies HMMs to opsins and olfactory receptors within Class A and to Classes B, C, frizzled, and smoothened.

Scientific Applications:

• Pharmacology and drug discovery: Predicts GPCR–G-protein coupling to inform target selection and anticipate signaling outcomes relevant to drug design.
• Functional annotation of GPCRs: Infers coupling selectivity to aid elucidation of GPCR subtype signaling mechanisms and functional roles.

Methodology:

Combines support vector machine (SVM) algorithms and hidden Markov models (HMMs); selects quantitative features from entire structural segments of ligands, GPCRs, and G-proteins to form feature vectors; uses a hierarchical SVM for Class A GPCRs and HMMs for opsins, olfactory receptors (Class A) and for Classes B, C, frizzled, and smoothened.