Explainable machine learning and deep learning models for predicting TAS2R-bitter molecule interactions.

Area of Science:

• Computational chemistry
• Pharmacology
• Bioinformatics

Background:

• Bitterness perception is mediated by taste receptor type 2 (TAS2R) G protein-coupled receptors.
• TAS2Rs have roles beyond taste, impacting various physiological functions and diseases.
• Predicting ligand-TAS2R interactions is crucial for taste perception and drug design.

Purpose of the Study:

• To develop explainable machine learning (ML) and deep learning (DL) models for predicting bitter molecule-TAS2R interactions.
• To leverage experimentally validated data for model training.
• To enhance understanding of bitter compound characteristics and TAS2R targeting.

Main Methods:

• Utilized traditional ML and DL approaches.
• Trained models on experimentally validated data.
• Integrated models for enhanced explainability and interpretation.

Main Results:

• Developed high-performance and applicable ML and DL models.
• Demonstrated synergistic integration of models for improved explainability.
• Facilitated interpretation of results regarding bitter compound-receptor interactions.

Conclusions:

• The developed models offer a powerful tool for predicting bitterant-TAS2R interactions.
• These models can guide the design of novel bitter compounds targeting specific TAS2Rs.
• The research advances understanding in taste perception and drug design.