design of novel quinazolines as potent anti-cancer compounds by coupling qsar and docking methods

Abstract: the main aim of this work is designing new derivatives of quinazolines that have desirable anti-cancer properties by coupling quantitative structure_activity relationship (qsar) and molecular docking methodes. quinazolines are a class of heterocyclic compounds that have unique properties in the field of medicinal chemistry, especially in treatment of cancer. in this study, a data series containing 48 molecules of quinazoline derivatives was selected from several valid sources [1-4]. the interaction of the molecules with the 4wkq protein was investigated during the molecular docking process and the best conformer of the data set molecules was used for qsar study. dragon software was used to calculate 1497 descriptors for each data set molecules. after evaluation of the calculated descriptors, 366 descriptors were remained for further investigation. the compounds were divided into training (39 molecules) and test (9 molecules) sets, where the training set was used for modeling process and the test set was used to evaluate the models. qsar models were developed by the stepwise-mlr, ga-ann, gsa-mlr and gsa-ann methods. the stepwise-mlr and ga-ann models were identified as invalid models, because their evaluation results were not satisfactory. the nonlinear gsa-ann model was recognized as the best obtained model, which includes eight molecular descriptors, ggl6, mats5p, gats2p, g(n...cl), rdf030p, mor23e, ish and h5u. the qsar model exhibited good statistical values for training (r2train=0.874) and test (r2test=0.866) sets. hence, it can claim that the gravitational search algorithm (gsa) is a more powerful descriptor selection method than others [5, 6]. finally, the model obtained by this method was used to predict the anticancer activity of new candidates of quinazoline compounds.