The main objective of this study is to develop 2D- and 3D-QSAR statistical models for a series consisting of 28 molecules. The authors started with a 2D study based on principal component analysis (PCA), multiple linear regression (RLM), and nonlinear regression (RNLM). The models were developed using 28 molecules with a pIC50 between 5.70 and 6.70. Then they applied 3D-QSAR analysis based on the partial least squares (PLS) method. For 3D-QSAR, they used the molecular field comparative analysis (CoMFA) and comparative molecular similarity index (CoMSIA) methods. For this analysis, they have worked on a training set of 24 compounds, which then give acceptable and reliable values of Q2 (0.791 and 0.538, respectively) and R2 (0.974 and 0.98, respectively). To determine the quantitative 3D-QSAR, the interpretations were based on the contour maps which are produced by the CoMFA and CoMSIA models. In addition, molecular docking is also one of the most important methods for confirming the binding interactions of predicted molecules with their receptors.
2D- and 3D-QSAR, molecular docking, and virtual screening of pyrido[2,3-d]pyrimidin-7-one-based CDK4 inhibitors