Machine learning models to predict ligand binding affinity for the orexin 1 receptor

Area of Science:

• Neuroscience
• Pharmacology
• Computational Chemistry

Background:

• The orexin 1 receptor (OX1R) is a G-protein coupled receptor implicated in various physiological processes.
• Selective OX1R antagonists show promise for behavioral disorders like drug seeking and overeating.
• There is a clinical need for novel selective OX1R antagonists due to the lack of approved drugs.

Purpose of the Study:

• To develop highly predictive quantitative structure-activity relationship (QSAR) models for orexin 1 receptor (OX1R) ligands.
• To identify novel OX1R ligands, including potential drug candidates, using virtual screening.
• To establish a robust QSAR modeling framework for future drug discovery efforts targeting OX1R.

Main Methods:

• A comprehensive dataset of over 1300 OX1R ligands was curated using stringent criteria.
• Random forest machine learning algorithm with optimized hyperparameters and 12 selected 2D molecular descriptors was employed.
• Recursive feature elimination with 5-fold cross-validation was used for descriptor selection and model validation.
• Virtual screening of the DrugBank database was performed using the developed QSAR model.
• Potential OX1R ligands were confirmed using radiolabeled OX1R binding assays.

Main Results:

• Highly predictive QSAR models for OX1R ligands were successfully developed.
• The QSAR model demonstrated high predictivity, validated by external test sets and enrichment studies.
• Virtual screening identified isavuconazole and cabozantinib, both FDA-approved drugs, as potential OX1R ligands.
• Binding assays confirmed the interaction of isavuconazole and cabozantinib with the OX1R.

Conclusions:

• This study presents the first highly predictive QSAR models for a large, diverse dataset of OX1R ligands.
• The developed QSAR models are valuable tools for the discovery and design of novel OX1R-targeting compounds.
• The identification of FDA-approved drugs as potential OX1R ligands opens new avenues for therapeutic development.