An Exploration of the Antiproliferative Potential of Chalcones and Dihydropyrazole Derivatives in Prostate Cancer via Androgen Receptor: Combined QSAR, Machine Learning, and Molecular Docking Techniques

In this study, the antiproliferative activities of some chalcones and dihydropyrazole derivatives in prostate cancer were investigated via the androgen receptor using QSAR, machine learning, and molecular docking techniques. A total of 30 dichloro substituted chalcones and dihydropyrazole derivatives were collected from the literature and optimized using density functional theory. Genetic function approximation was employed for model development. The developed model was thoroughly validated. Its generalization and predictive capacities were improved with the extreme learning machine (ELM) algorithm. Molecular docking and drug-likeness screening of the compounds were carefully performed. A reduction in the negative coefficient of the descriptor and an increase in the positive coefficient of the descriptor improved bioactivity. An R2 pred value of 0.737 showed a strong correlation between the experimental and predicted activities. A correlation coefficient of 0.8305 for R2demonstrated the predictability of the model. The ELM-Sine model showed an improvement of 66.7% and 8.3% in QSAR and ELM-Sig models, respectively. Molecular docking showed the chalcones and dihydropyrazole derivatives to be promising anti-prostate cancer agents, with pi-pi stacking and hydrogen bond interactions favoring the inhibition of the androgen receptor. The lead drugs are drug-like and novel anti-prostate cancer agents.