GPCRHMM

GPCRHMM identifies G protein-coupled receptors (GPCRs) in protein sequences using a hidden Markov model tuned to GPCR structural and compositional features, including seven-transmembrane topology with an extracellular N-terminus and a cytosolic C-terminus.

Key Features:

• Customized Hidden Markov Model: Trained on an extensive, diverse GPCR dataset to capture loop length patterns and variations in amino acid composition across cytosolic loops, extracellular loops, and membrane-spanning segments.
• Seven-transmembrane topology: Targets the common GPCR motif of seven transmembrane helices with extracellular N-terminus and cytosolic C-terminus.
• Biological context: Designed for receptors activated by diverse ligands including hormones, odorants, peptides, and proteins.
• Benchmark performance: In cross-validation, achieved an error rate nearly half that of profile HMMs and approximately 15% higher sensitivity than the best transmembrane predictors while maintaining comparable false positive rates.
• Novel GPCR detection: Used to search five proteomes and identified 120 candidate novel GPCR sequences, including 102 in Caenorhabditis elegans, four in humans, and seven in mice.
• Functional domain associations: Many predictions (65) were linked to Pfam domains of unknown function.

Scientific Applications:

• Genomic discovery: Identification of previously unannotated GPCRs across proteomes to expand the known GPCR superfamily.
• Annotation and classification refinement: Discrimination and rejection of sequences that do not match GPCR characteristics, such as certain arthropod-specific odorant receptors, to improve functional annotation and classification.

Methodology:

Configure and train a hidden Markov model using analyses of loop length patterns and regional amino acid composition from a diverse GPCR dataset, and evaluate performance via cross-validation against profile HMMs and transmembrane predictors.