Applications of machine learning in GPCR bioactive ligand discovery

Area of Science:

• Pharmacology
• Computational Chemistry
• Bioinformatics

Background:

• G protein-coupled receptors (GPCRs) are a major class of drug targets, with 475 FDA-approved drugs.
• The pharmaceutical industry invests heavily in identifying potent GPCR ligands.
• Vast chemical databases necessitate efficient screening methods for drug discovery.

Purpose of the Study:

• To provide a cheminformatics overview of machine learning (ML) applications in GPCR drug discovery.
• To investigate ML's role in both ligand-based and structure-based virtual screening.
• To present a case study on predicting agonists for an olfactory receptor (OR1G1).

Main Methods:

• Review of machine learning applications in GPCR drug discovery.
• Cheminformatics analysis.
• Comparison of four classical ML methods for agonist prediction in a case study involving OR1G1.

Main Results:

• Machine learning methods are crucial for navigating large chemical spaces in GPCR drug discovery.
• ML approaches are applicable to various stages of the drug discovery pipeline.
• The case study demonstrated the utility of ML in predicting agonists for olfactory receptors.

Conclusions:

• Machine learning is an essential tool for efficient GPCR drug discovery.
• ML facilitates the identification of lead compounds from extensive chemical libraries.
• Further application of ML, particularly in olfactory receptor research, holds significant promise.