| Accurate prediction of protein-ligand affinity can provide useful help in the drug design.Although the number of prediction methods for evaluating affinity is large,the accuracy of those methods is often low.For improving the efficiency of the drug design,it is necessary to compare those existing prediction methods and to develop a highly accurate method for evaluating the protein-ligand affinity.In this work,trypsin-ligand complexes are chosen as the research object.Based on the Ploarizable Bond-dipole-based Force Field(PBFF)developed in our group,a special method for calculating the affinity of the trypsin-ligand complexes was developed.The comparison between our method with other methods(MM-PB/SA,MM-GB/SA,and X-Score)was performed.Research results show:1.In the PBFF method,the combination of a dispered charge distribution scheme for the ligand and a concentrated distribution scheme for the protein can improve the accuracy of the prediction of trypsin-ligand binding affinity.2.Compared with MM-PB/SA,MM-GB/SA,and X-Score methods,the results calculated by the PBFF method makes a good agreement with the experimental data.For MM-PB/SA,MM-GB/SA,X-Score,and PBFF methods,the correlation coefficients are 0.16,0.26,0.20,and 0.63 respectively;the root mean square error values are 1.83,20.12,22.94,and 1.02 kcal/mol;the mean absolute deviation error values are 1.47,19.72,22.60,and 0.83 kcal/mol;the mean relative deviation values are 0.35,3.76,4.41,0.19.It is hoped that the above research results can help to develop a new high-precision method for the serine protease and its ligand complex. |
