Meta^QM: Exploring the Role of QM Calculations in Drug Metabolism Prediction

Area of Science:

• Computational chemistry
• Drug discovery
• Pharmacokinetics

Background:

• Predicting xenobiotic metabolism is vital for identifying safe and effective drug candidates early in development.
• Poor ADMET properties are a major reason for drug candidate failure.
• In silico metabolism modeling aids in designing better compounds.

Purpose of the Study:

• To develop MetaQM, a computational tool utilizing quantum chemical descriptors for predicting xenobiotic metabolism.
• To predict the occurrence of metabolic reactions (MetaclassQM) and the site of metabolism (MetaspotQM).

Main Methods:

• Trained random forest classifiers on the MetaQSAR database (3788 reactions).
• Employed quantum chemical descriptors (PM7 and DFT) including physicochemical, constitutional, and stereo-electronic features.
• Evaluated model performance for reaction classification and site prediction.

Main Results:

• DFT descriptors improved MetaclassQM classification by 10% (class) and 8.6% (subclass) compared to PM7.
• Both DFT and PM7 descriptors showed similar performance for MetaspotQM site of metabolism prediction.
• Quantum descriptors enhance metabolic reaction classification accuracy.

Conclusions:

• DFT descriptors offer superior performance for classifying metabolic reactions.
• Simpler computational methods are adequate for predicting metabolic sites.
• Quantum descriptors provide a valuable balance of accuracy and computational efficiency in metabolism modeling.