Machine Learning-Enhanced Quantitative Structure-Activity Relationship Modeling for DNA Polymerase Inhibitor Discovery: Algorithm Development and Validation | JoVE Visualize

Area of Science:

Computational chemistry and cheminformatics
Drug discovery and development
Molecular biology and oncology

Background:

Cisplatin resistance in cancer therapy is a major challenge, often mediated by translesion DNA synthesis involving human DNA polymerase η (hpol η).
Existing small-molecule inhibitors of hpol η, like PNR-7-02, often lack the potency and specificity required to overcome chemoresistance.
The vast chemical space necessitates advanced computational methods, such as machine learning (ML)-enhanced quantitative structure-activity relationship (QSAR) modeling, for efficient drug discovery.

Purpose of the Study:

To develop and validate ML-augmented QSAR models for predicting hpol η inhibition.
To accelerate the discovery of potent and selective indole thio-barbituric acid analogs as hpol η inhibitors.
To identify novel therapeutic strategies for overcoming cisplatin resistance in cancer treatment.

Main Methods:

A library of 85 indole thio-barbituric acid analogs with known hpol η inhibition data was curated.
220 molecular descriptors (1D-4D) were computed, and 17 ML algorithms were trained and validated using 80% of the data.
Hyperparameter optimization and 5-fold cross-validation were employed to ensure model robustness and performance evaluation using 14 metrics.

Main Results:

Ensemble ML methods, particularly random forest, demonstrated exceptional predictive performance (R² > 0.9998) for hpol η inhibition.
Shapley additive explanations identified electronic properties, lipophilicity, and topological atomic distances as key predictors.
Nonlinear relationships between molecular descriptors and inhibitory activity were highlighted, with linear models showing higher errors.

Conclusions:

ML-integrated QSAR modeling offers a powerful and interpretable framework for optimizing hpol η inhibitors.
This approach significantly accelerates the identification of potent and selective compounds to combat cisplatin resistance.
The study advances precision oncology by providing a strategy to overcome a critical challenge in cancer therapy.