Developing muscarinic receptor M1 classification models utilizing transfer learning and generative AI techniques.

Area of Science:

• Pharmacology
• Computational Chemistry
• Machine Learning

Background:

• Muscarinic receptor subtype 1 (M1) is a G protein-coupled receptor (GPCR) crucial for treating peripheral neuropathy, COPD, and cognitive disorders.
• Identifying M1-interacting compounds is vital for rational drug design.
• Publicly available bioactivity data is a valuable resource for drug discovery.

Purpose of the Study:

• To develop machine learning-based classification models for the M1 receptor.
• To address the challenge of imbalanced datasets common in bioactivity data.
• To enhance the screening of chemical databases for potential M1-targeting drugs.

Main Methods:

• Utilized publicly available bioactivity data for M1 receptor.
• Investigated transfer learning strategies to improve model performance.
• Employed generative models for oversampling the inactive compound class.
• Developed M1 classification models using machine learning algorithms.

Main Results:

• Successfully developed machine learning models for M1 receptor classification.
• Demonstrated that transfer learning and generative oversampling reduce misclassification of inactive compounds.
• Observed improved classification performance for M1 and other GPCR targets.
• Validated the effectiveness of the developed models on imbalanced datasets.

Conclusions:

• The developed M1 classification models enable rapid screening of large chemical libraries.
• These models advance the rational drug design process for M1-related disorders.
• The strategies employed are applicable to improving models for other GPCR targets.
• This work facilitates efficient drug discovery for significant therapeutic areas.