MOLECULAR DYNAMICS GUIDED 4D QSAR STUDIES ON CHROMENONE BASED DNA-DEPENDENT PROTEIN KINASE INHIBITORS

摘要

Objective: Developing 4D QSAR model for chromenones as DNA-protein kinase inhibitors with LQTAgrid program and MATLAB with PLS Tool box. Methods: 2D structures of forty five chromenones and pyrimido [2,1-a] isoquinolin-4-ones were drawn in Marvin application and subsequently converted to 3D format. With PRODRG server coordinate and topology files were generated. Molecular dynamics simulation for 20 ps was carried out in Gromacs4.6.3. Conformation ensemble for each molecule was aligned with the most active compound 22. Interaction energy descriptors were computed by placing NH 3 + probe with LQTA grid program. 4D QSAR model was built-in MATLAB workspace with PLS Toolbox. Results: 4D QSAR model developed with PLS regression showed root mean square error 0.188741, root mean square error predicted 0.575649, prediction bias -0.0221941, R 2 calculated for training set 0.834502. The model built could predict the DNA-PK inhibitory activity for analogous compounds with reasonably good predictive ability. The Leenard Jones interaction descriptors were found important in the model suggesting the importance of 3D steric features in the molecule. These facts could be exploited in designing newer DNK-PK inhibitors with improved activity. Conclusion: 4D QSAR model was developed with good squared correlation coefficient of 0.834502. Few Leenard Jones descriptors were found contributing to the model. The 4D QSAR model could be used in designing more active DNK-PK inhibitors. Keywords: Chromenones, 4D QSAR, DNA-Protein kinase inhibitors, LQTA grid, MATLAB, PLS regression.