Integrated Multi-Class Classification and Prediction of GPCR Allosteric Modulators by Machine Learning Intelligence.

Area of Science:

• Pharmacology
• Computational Chemistry
• Bioinformatics

Background:

• G-protein-coupled receptors (GPCRs) are a vast family of cell surface receptors crucial for cellular communication.
• Allosteric modulation of GPCRs is a key strategy for developing targeted therapies and understanding signaling pathways.
• Accurately classifying GPCR modulators across diverse subtypes is challenging but essential for drug development.

Purpose of the Study:

• To develop and evaluate machine learning models for the multi-class classification of GPCR allosteric modulators.
• To differentiate between modulators of GPCR families A, B, and C, and random compounds.
• To establish efficient computational tools for identifying novel GPCR allosteric modulators.

Main Methods:

• Utilized a dataset of 34,434 compounds, including known GPCR allosteric modulators and random compounds.
• Trained six machine learning models: Support Vector Machine, Naïve Bayes, Decision Tree, Random Forest, Logistic Regression, and Multilayer Perceptron.
• Employed various feature sets: molecular descriptors, Atom-pair fingerprints, MACCS fingerprints, and ECFP6 fingerprints.

Main Results:

• Investigated the performance of different machine learning models with various feature combinations for an 11-class classification task.
• Demonstrated the feasibility of using machine learning for accurate classification of GPCR allosteric modulators.
• Identified effective feature sets and model architectures for distinguishing GPCR modulator classes.

Conclusions:

• The developed machine learning models provide a robust and efficient method for classifying GPCR allosteric modulators.
• These tools can significantly aid in the discovery and development pipeline of novel GPCR-targeted therapeutics.
• This work represents a novel approach to the multi-class classification of GPCR allosteric modulators.