SHAP

SHAP explains contributions of chemical substructures to predicted metabolic stability of compounds using SHapley Additive exPlanations.

Key Features:

• Predictive Modeling: Utilizes multiple predictive models trained and validated on ChEMBL data to assess compound metabolic stability.
• SHAP Value Analysis: Computes SHAP (SHapley Additive exPlanations) values to quantify the contribution of specific substructures to model predictions.
• Structural Modification Insights: Compares SHAP patterns among the most similar compounds in the dataset to highlight privileged and unfavorable chemical moieties for stability optimization.
• Application in Drug Design: Provides substructure-level information relevant to ligand design and pharmacokinetic property optimization.

Scientific Applications:

• Metabolic Stability Optimization: Evaluates how substructures influence predicted metabolic stability, affecting the duration of compound activity in biological systems.
• Drug Design Enhancement: Informs optimization of physicochemical and structural properties during medicinal chemistry campaigns.
• Research and Development Support: Enables analysis-driven selection and modification of compound candidates in early-stage drug discovery.

Methodology:

Predictive models are trained and validated on ChEMBL data, SHAP values are computed to attribute substructure contributions to stability predictions, and analogous-compound analysis compares SHAP patterns among the most similar dataset compounds.