PIC50PRED: A PIC50 PREDICTION TOOL FOR 5-ALPHA REDUCTASE ENZYME
Asian Journal of Pharmaceutical and Clinical Research · 2016
ABSTRACT Objectives: Prostate cancer is a major health burden all over the world. 5-alpha reductase (5AR) enzyme is a significant drug target for prostate cancer. Identification of drug targets and their inhibitors are a challenging task in drug designing. The prediction of potential inhibitors against 5AR may help in designing effective drugs against prostate cancer. Methods: The compounds having proven inhibitory action against 5AR in experimental settings were trained and tested to build two-dimensional quantitative structure-activity relationship (2D QSAR) models based on molecular descriptors. The molecular descriptors were extracted from E-Dragon 1.0 program. The 2D QSAR prediction models were built using linear regression and least median of squares using Weka 3.7. The optimized 2D QSAR models were implemented in a web-based server by employing XAMPP package and using hyper processed language (PHP) as a scripting language. Results: The 2D QSAR models were built using molecular descriptors and achieved a positive correlation of 0.69 (r) and 0.46 (r) between predicted and actual pIC50 from linear regression and least square of median, respectively. Conclusion: In silico QSAR modeling along with machine learning techniques seems to be a promising approach for prediction of novel 5AR inhibitors. To serve the scientific community, a web server “pIC50 Pred” has been developed which allows the prediction of pIC50 value of any novel compounds thought to have 5AR inhibitory activity before jumping into in vitro experimental assays. Availability: The prediction tool is freely available at http://www.biopred.org. Keywords: 5-alpha reductase, Two-dimensional quantitative structure-activity relationship, pIC50, Weka, Linear regression, Least median of squares.