Evaluation Of XO Method Based On CASF2016?MCL-1 And CDK2 Test Set

Accurately calculating the interaction between proteins and ligands has always been a significant things in theoretical computational.If the theoretical calculation method can accurately calculate the binding energy of proteins and ligands,it is great significance for drug design.For this purpose,researchers have done a lot of research work.In computer-aided drug design,the most widely used method for assessing binding affinity is the empirical scoring function in molecular docking.It is well known that the scoring function based on the molecular force field is not highly accurate due to the heavily empirical parameters.In this case,people introduced the method of quantum mechanics to evaluate binding affinity.Although a high theoretical level can bring accurate calculation results,the calculation cost will also increase.Therefore,how to find a balance between calculation accuracy and calculation efficiency is particularly important.For this purpose,this paper used a strategy based on the "divide and conquer" method and first-principles quantum mechanics to balance the calculation accuracy and calculation efficiency.That is,the combination of coB97X-D density functional theory and XO method is used to calculate protein-ligand binding affinity.The main purpose of this article is to test the method's ability to evaluate protein-ligand binding affinity.For this purpose,two virtual drug design scenarios were simulated and a large number of tests were conducted on several different test sets.The first is a universal test on the CASF2016 test set for the lead compound discovery scenario.The test set contains a total of 57 different types of proteins,including most of the current popular targets in pharmaceutical research.Each target contains 5 inhibitors,with diverse pharmacophores and inhibitor structures.It not only guarantees the richness of the test set,but also the mathematical statistical significance.By calculating each target and sorting the calculated binding energy,the average Spearman correlation coefficient is 0.68,which is significantly better than the results obtained from 34 scoring functions reported in the literature,which fully proves the advantages of this method.The second is the optimization test of the lead compound.In the process of drug development,it is necessary to optimize the modification of a lead compound in order to obtain a more active and specific inhibitor.Therefore,theoretical calculation methods are required to be able to distinguish the difference in activity between two compounds with the same skeleton structure.This article uses the two test sets of MCL-1 and CDK2.In both test sets,the inhibitors each contain a common skeleton,have large amounts of data(including 66 and 76 inhibitors,respectively),and have widely distributed experimental activity values.The Pearson correlation coefficient of relative binding energy and biological activity calculated on the MCL-1 test set was 0.81,and the Pearson correlation coefficient of relative binding energy and biological activity calculated on the CDK2 test set was 0.86.The above results show that this strategy can provide an effective theoretical tool for the optimization of lead compounds.