EMCBOW-GPCR

EMCBOW-GPCR predicts G Protein-Coupled Receptors (GPCRs) from primary protein sequences using natural language processing-based word embeddings and machine learning to enable accurate GPCR identification.

Key Features:

• Natural Language Processing (NLP) Integration: Applies NLP techniques to model protein sequence analysis analogously to text data processing.
• Feature Extraction Models: Uses three distinct word-embedding models alongside a bag-of-words model to extract original features from GPCR primary sequences.
• Deep Learning for Feature Enhancement: Employs deep learning algorithms to refine feature representations and reduce dimensionality.
• Extreme Gradient Boosting (XGBoost): Classifies the processed features using Extreme Gradient Boosting for GPCR prediction.
• Comparative Predictive Performance: Demonstrates performance that outperforms existing state-of-the-art methods on overall prediction metrics.

Scientific Applications:

• GPCR Identification: Enables computational identification of G Protein-Coupled Receptors from sequence data as an alternative to experimental screening.
• Drug Discovery and Development: Supports studies of receptor–ligand interactions and target identification in therapeutic development.
• Membrane Protein Research: Facilitates analysis and screening of membrane protein sequence datasets.

Methodology:

GPCR primary sequences are represented using three word-embedding models and a bag-of-words model, processed by deep learning to extract, refine, and reduce feature dimensionality, and classified with Extreme Gradient Boosting (XGBoost).