Monte Carlo based modelling approach for designing and predicting cytotoxicity of 2-phenylindole derivatives against breast cancer cell line MCF7

Abstract

Breast cancer is one of the leading causes of cancers among the variety of cancers in woman all over the world. Compounds with phenylindole scaffold were found to execute promising cytotoxicity against breast cancer cell line MCF7. In the present study, a Monte Carlo based QSAR analysis was performed on a dataset containing 102 phenylindoles in order to accelerate the efforts to find out better cytotoxic phenylindoles against MCF7 cell line. The statistical qualities of the generated models were found to be quite good as far as the internal and external validation were concerned. The best models from each split (Split 1: R2 = 0.6944, Q2 = 0.6495; Split 2: R2 = 0.8202, Q2 = 0.7998; Split 3: R2 = 0.8603, Q2 = 0.8357) for the test set were selected and Y-scrambling test and applicability domain analysis were also performed to ensure the robustness of these models. Among these models, model from split 3 obtained by using hybrid descriptors (combination of SMILES and HSG with 0ECk connectivity) was used to identify and classify the structural attributes as promoters as well as hinderers of cytotoxicity for these 2-phenylindole derivatives. Results from the analysis were further used to design and predict some probable new 2-phenylindole derivatives having promising cytotoxicity (IC50 < 55 nM) against MCF7.