GPCR-CA

GPCR-CA identifies and classifies G-protein-coupled receptors (GPCRs) from protein sequences to determine GPCR membership and assign one of six functional classes (rhodopsin-like, secretin-like, metabotrophic/glutamate/pheromone, fungal pheromone, cAMP receptor, and frizzled/smoothened family) for comparative and evolutionary pharmacology.

Key Features:

• Two-Layer Prediction Engine: A first layer discriminates GPCR versus non-GPCR sequences and a second layer assigns identified GPCRs to one of six functional classes.
• Cellular Automaton (CA) Methodology: Cellular Automata are used to generate CA images from protein sequences to reveal hidden pattern features.
• Feature Quantification: CA image patterns are quantified using gray-level co-occurrence matrix factors and proteins are represented by pseudo amino acid composition.
• Performance and Validation: Reported success rates exceed 91% for GPCR identification and 83% for functional class classification, evaluated by jackknife cross-validation.
• Benchmark Dataset: Validation used a stringent benchmark dataset constructed so that no protein pair within the same subset shares ≥40% sequence identity.

Scientific Applications:

• Functional Annotation: Classifying uncharacterized protein sequences as GPCRs and assigning their functional class for molecular biology studies.
• Drug Discovery and Development: Informing target identification and comparative pharmacology analyses by providing GPCR membership and class information.
• Comparative and Evolutionary Pharmacology: Supporting analyses of evolutionary relationships and functional diversification among GPCR families.

Methodology:

Uses a two-layer prediction approach; generates Cellular Automata (CA) images from sequences; quantifies CA patterns with gray-level co-occurrence matrix factors; represents proteins by pseudo amino acid composition; and evaluates performance by jackknife cross-validation on a dataset with no within-subset pairs ≥40% sequence identity.