A 3D-QSAR Analysis of CDK2 Inhibitors Using FMO Calculations and PLS Regression.

We report a three-dimensional quantitative structure-activity relationship (3D-QSAR) analysis of CDK2 inhibitors using fragment molecular orbital (FMO) calculations and partial least squares (PLS) regression. In our analysis, fragment binding energies of individual amino acids and fragment binding energy of a single ligand in a protein-ligand complex are evaluated by FMO calculations and used as descriptors in PLS regression to estimate biological activities of the ligands. The analysis was applied to the system of CDK2 protein and its inhibitors and the effectiveness of the method was tested. Application of the 3D-QSAR model demonstrated that it offered good predictive ability and was able to predict not only biological activity of ligands but also identify important amino acid residues which could be targeted in order to improve ligand activity.