Interpretable QSAR and Complementary Docking for PARP1 Inhibitor Prioritization: Reliability Stratification and...

Area of Science:

• Medicinal Chemistry
• Cheminformatics
• Computational Drug Discovery

Background:

• Poly(ADP-ribose) polymerase 1 (PARP1) is a key therapeutic target in DNA repair-deficient cancers.
• Challenges in PARP1 inhibitor discovery include scaffold convergence, tolerability issues, and acquired resistance.
• Development of novel, effective PARP1 inhibitors is crucial for advancing cancer treatment.

Purpose of the Study:

• To create an interpretable and reliable cheminformatics workflow for prioritizing PARP1 inhibitors.
• To integrate quantitative structure-activity relationship (QSAR) modeling with structure-based drug design.
• To facilitate the identification of novel PARP1 inhibitors with improved potency and drug-like properties.

Main Methods:

• A curated dataset of 3339 PARP1 inhibitors from ChEMBL was utilized.
• Machine learning models, including a stacked ensemble, were trained using RDKit descriptors and Avalon fingerprints.
• Feature selection, model interpretation (SHAP, permutation importance), external validation via PubChem similarity expansion, and molecular docking (AutoDock Vina, SwissDock) were employed.

Main Results:

• The stacked ensemble model demonstrated strong predictive performance (test R² = 0.723, RMSE = 0.610 pIC₅₀ units).
• External validation confirmed the model's utility, with a positive association between predicted and experimental activities (R² = 0.124, Pearson r = 0.479).
• Three promising drug candidates (CID 168873053, CID 175154210, CID 172894737) were identified through complementary docking analysis.

Conclusions:

• The developed workflow offers a transparent and practical framework for prioritizing PARP1 inhibitors.
• The integrated approach of QSAR, explainability, external corroboration, and docking effectively supports the generation of lead compounds.
• This strategy aids in accelerating the discovery of novel PARP1 inhibitors for experimental validation and potential therapeutic applications.