QSAR Modeling for Predicting IC₅₀ and GI₅₀ Values for Human Cell Lines Used in Toxicological Studies.

Area of Science:

• Pharmacology
• Toxicology
• Computational Chemistry

Background:

• Cytotoxicity assessment is vital for preclinical drug development, guiding the evaluation of new pharmaceutical agents.
• Human cell lines enable the analysis of general and organ-specific toxicity, reducing development time and costs.
• Predicting cytotoxic IC50 and GI50 values helps assess the therapeutic window early in development.

Purpose of the Study:

• To develop Quantitative Structure-Activity Relationship (QSAR) models for predicting cytotoxicity in human cell lines.
• To implement these models into a user-friendly web application for broader accessibility.

Main Methods:

• Selected 20 human cell lines (10 non-tumor, 10 tumor) commonly used in toxicology.
• Utilized GUSAR software with atom-centric electrotopological QNA and substructural MNA descriptors.
• Employed the RBF-SCR algorithm for QSAR model training and 5-fold cross-validation for model selection.

Main Results:

• Developed QSAR models with predictive performance (mean R² = 0.691, mean RMSE = 0.584).
• Selected best-performing models exceeding R² > 0.5 and RMSE < 0.8.
• Integrated the models into the CLC-Pred 2.0 web application for qualitative cytotoxicity prediction.

Conclusions:

• The developed QSAR models provide a reliable method for predicting human cell line cytotoxicity.
• CLC-Pred 2.0 offers a valuable, freely available tool for early-stage drug safety assessment in preclinical toxicology.
• This approach can accelerate drug development by facilitating early identification of potential toxic compounds.