Molecular docking approach has been extensively used to predict the ligand’s binding conformation in the binding pocket of protein. However, its prediction accuracy is still limited and highly dependent on target protein-ligand complexes. In this study, we investigated the effects of ligand torsion number, ligand hydrophobicity, and binding-site hydrophobicity on the prediction accuracy of Autodock, a popular molecular docking tool, combinatorially as well as respectively. A clear understanding of how these properties affect the prediction accuracy was observed when these properties were studied combinatorially rather than individually. The combination of low ligand torsion number-hydrophilic ligand-hydrophobic binding site provided the best prediction accuracy while the high ligand torsion number-hydrophilic ligand-hydrophobic binding pocket combination showed the least prediction accuracy. This study allowed us to determine the molecular properties of complex, showing relatively higher or low prediction accuracy and can be employed as a reference in the molecular docking studies using Autodock.

分子对接方法已广泛用于预测蛋白质结合口袋中的配体结合构象。但是，其预测准确性仍然有限，并且高度依赖于靶蛋白-配体复合物。在这项研究中，我们分别研究了配体扭转数，配体疏水性和结合位点疏水性对流行的分子对接工具Autodock的预测准确度的影响，以及组合影响。通过组合而不是单独研究这些属性，可以清楚地了解这些属性如何影响预测精度。低配体扭转数-亲水性配体-疏水结合位点的组合提供了最佳的预测准确性，而高配体扭转数-亲水性配体-疏水结合囊结合的组合显示了最低的预测准确性。这项研究使我们能够确定复合物的分子特性，显示出相对较高或较低的预测准确度，并且可以用作使用Autodock进行分子对接研究的参考。