A Linear and Nonlinear QSAR Analysis of Benzimidazole Derivative XY123 in Prostate Cancer Treatment

Background: Metastatic Castration-Resistant Prostate Cancer (mCRPC) represents a critical challenge in current prostate cancer treatment. Benzimidazole Derivative XY123 has emerged as a novel inhibitor for its treatment. Objective: This study aims to establish a robust Quantitative Structure-Activity Relationship (QSAR) model for predicting the activity of Benzimidazole Derivative XY123 derivatives, aiding the development of novel anti-prostate cancer drugs. Methods: Utilizing CODESSA software, descriptors were computed based on various moieties of Benzimidazole Derivative XY123 derivatives. Multiple linear regression models were constructed, and both linear and nonlinear QSAR models were developed using heuristics and gene expression programming. Results: The linear model with two descriptors demonstrated the best predictive capacity for inhibitor activity, while the nonlinear model generated through Gene Expression Programming (GEP) exhibited correlation coefficients of 0.83 and 0.82 for the training and test sets, respectively. The average errors were 0.03 and 0.05, indicating the stability and the improved predictive ability of the nonlinear model. Conclusion: The QSAR linear model has an advantage over the nonlinear model in optimizing Benzimidazole Derivative XY123, providing a direction for the development of effective drugs for mCRPC treatment.