Metabolite Identification Data in Drug Discovery, Part 2: Site-of-Metabolism Annotation, Analysis, and Exploration...

Area of Science:

• Drug discovery and development
• Medicinal chemistry
• Pharmacokinetics

Background:

• Accurate prediction of sites of metabolism (SoMs) is vital for designing safe and effective small molecules.
• Limited availability of annotated SoM data impedes the progress of data-driven prediction methods, including machine learning.
• Existing SoM datasets often lack comprehensive characterization and do not fully account for experimental data uncertainty.

Purpose of the Study:

• To comprehensively characterize SoM data from human hepatocyte assays.
• To develop and validate a novel strategy for SoM annotation that incorporates experimental uncertainty.
• To enhance the accuracy of SoM prediction models using newly generated and analyzed metabolism data.

Main Methods:

• Acquisition and characterization of SoM data from human hepatocyte assays.
• Development of a new SoM annotation strategy addressing data uncertainty.
• Entropy analysis of SoM annotations to understand data complexity.
• Evaluation of the impact of new data on SoM prediction model performance.

Main Results:

• A substantial dataset of SoM annotations was generated and characterized.
• The novel annotation strategy effectively handles uncertainty in experimental metabolism data.
• Entropy analysis revealed complexities in interpreting metabolism data.
• The newly available data significantly improved the performance of SoM prediction models.

Conclusions:

• The developed SoM annotation strategy enhances the reliability of metabolism data for computational modeling.
• The public release of this comprehensive SoM dataset will accelerate the advancement of machine learning-based metabolism prediction.
• This work provides a valuable resource for researchers in drug discovery aiming to optimize small molecule properties.