Estudo por docagem molecular de interações proteínas-ligantes utilizando diferentes ajustes de cargas

Abstract

Molecular docking is a technique widely used in the study of drug design in order to predict the binding modes of a ligand in a protein and its binding energy. The method is also widely used for virtual screening when databases of molecules with possible biological activity are docked to a receptor searching for new drugs. To increase the accuracy of the docking technique with the AutoDock 4.2 program, which has terms dependent on partial charges in the scoring function, this study combines different methods of assigning partial charges to receptors and ligands in order to analyze which combinations get a better prediction of docking energies and binding poses. The study combined the charges obtained by the AMBER03, Gasteiger, and GROMOS96 methods for the receptors and for the ligands the charges obtained with the AM1-BCC, CHARMM, Gasteiger, MMFF94, PM7, and RM1 methods. In total, 49 protein-ligand complexes were chosen to be re-docked using the eighteen combinations of charge assignment methods. The results show that the Gasteiger/RM1 charge assignment combination obtained the best R² among all combinations. However, it did not obtain the highest number of complexes with binding poses below the 2 A RMSD threshold. Despite being below average in the prediction of docking energy, the combination with the Gasteiger method assigned to the ligands obtained a consistent performance among the combinations with the charge methods for the receptors. In addition, combinations such as AMBER03/AM1-BCC and GROMOS96/MMFF94 obtained higher performance than the others in terms of success rates for ligands within the proposed 2 A RMSD threshold, thus being able to be chosen to be used in cases where experimental data is not known for the binding pose of the ligand.