Machine learning study: from the toxicity studies to tetrahydrocannabinol effects on Parkinson's disease

Abstract

Aim: Investigating molecules having toxicity and chemical similarity to find hit molecules. Methods: The machine learning (ML) model was developed to predict the arylhydrocarbon receptor activity of anti-Parkinson's and US FDA-approved drugs. The ML algorithm was a support vector machine, and the dataset was Tox21. Results: The ML model predicted apomorphine in anti-Parkinson's drugs and 73 molecules in FDA-approved drugs as active. The authors were curious if there is any molecule like apomorphine in these 73 molecules. A fingerprint similarity analysis of these molecules was conducted and found tetrahydrocannabinol (THC). Molecular docking studies of THC for dopamine receptor 1 (affinity = -8.2 kcal/mol) were performed. Conclusion: THC may affect dopamine receptors directly and could be useful for Parkinson's disease.