Transforming growth factor type 1 receptor (ALK5) is kinase associated with a wide variety of pathological processes, and inhibition of ALK5 is a good strategy to treat many kinds of cancer and fibrotic diseases. Recently, a series of compounds have been synthesized as ALK5 inhibitors. However, the study of their selectivity against other potential targets remains elusive. In this research, a data-set of ALK5 inhibitors were collected and studied based on the combination of 2D-QSAR, molecular docking and molecular dynamics simulation. The quality of QSAR models were assessed statistically by F, R2, and R2ADJ, proved to be credible. The cross-validations for the models (q2LOO = 0.571 and 0.629, respectively) showed their robustness, while the external validations (r2test = 0.703 and 0.764, respectively) showed their predictive power. Besides, the predicted binding free energy results calculated by MM/GBSA method were in accordance with the experimental data, and the van der Waals energy term was the factor that had the most significant impact on ligand binding. What is more, several important residues were found to significantly affect the binding affinity. Finally, based on our analyses above, a proposed series of molecules were designed.
Keywords: , Molecular Mechanics/Generalized Born Surface Area; , general Amber force field; , leave-one-out; , molecular dynamics; , partial least square analysis; , particle mesh Ewald; , quantitative structure–activity relationship; , root-mean-square deviation; , root-mean-square fluctuation; , transforming growth factor beta; , transforming growth factor beta type I receptor; 2D-QSAR; ALK5; MM/GBSA calculation; molecular dynamics simulation; small molecule inhibitor.