GPCRVS

GPCRVS employs machine learning to predict compound activity and interactions with G protein-coupled receptors (GPCRs), including peptide- and protein-binding GPCRs, as a decision support system for virtual screening.

Key Features:

• Activity Range Evaluation: GPCRVS assesses compounds in terms of their activity range against several GPCR targets.
• Pharmacological Effect Prediction: The system predicts the pharmacological effect compounds exert on the receptor.
• Binding Mode Prediction: GPCRVS predicts the binding mode compounds could demonstrate for different types and subtypes of GPCRs.
• Multiclass Classification: The system handles incomplete and fuzzy biological data through multiclass classification validated on ChEMBL and Google Patents-retrieved data sets for class B GPCRs and chemokine CC and CXC receptors.
• Comparative Analysis: Results of activity class assignment and binding affinity prediction are provided in comparison with predictions for known active ligands of each included GPCR.
• Compound Flexibility: GPCRVS evaluates compounds ranging from small molecules to short peptides provided in common chemical file formats.

Scientific Applications:

• GPCR Virtual Screening: GPCRVS enables efficient online assessment of compound activity against GPCR targets, including challenging peptide- and protein-binding GPCRs.
• Drug Design Support: The system supports decision-making in drug design by predicting compound activity, pharmacological effects, and binding modes across GPCR types and subtypes.
• Class B GPCR and Chemokine Receptor Analysis: Validated applications include prediction for class B GPCRs and chemokine CC and CXC receptors using data from ChEMBL and Google Patents.

Methodology:

GPCRVS uses machine learning-based multiclass classification trained on GPCR structures in active and inactive conformations together with their active ligands, with validation performed on ChEMBL and Google Patents-retrieved data sets.