CoMSiA model

Amin EA, Welsh WJ (2006) Highly predictive CoMFA and CoMSIA models for two series of stromelysin-1 (MMP-3) inhibitors elucidate SI and S1-S2 binding modes. J Chem Inf Model 46 1775-1783... 

Adamantan-1 -yl-quinoline-2-carboxylic acid alkylidene hydrazides 30 CoMFA and CoMSIA Models tested with 14 molecules CoMFA (R2 = 0.49) and CoMSIA (R2 = 0.49) Nayyar et al. (44)... 

Enoyl-acyl carrier protein reductase arylamide inhibitors were used to build CoMFA and CoMSIA models (tested with eight molecules,... 

Moreover, a final 3D-QSAR model vahdation was done using a prospective study with an external test set. The 82 compounds from the data set were used in a lead optimization project. A CoMFA model gave an (cross validated) value of 0.698 for four relevant PLS components and a conventional of 0.938 were obtained for those 82 compounds. The steric descriptors contributed 54% to the total variance, whereas the electrostatic field explained 46%. The CoMSIA model led to an (cross vahdated) value of 0.660 for five PLS components and a conventional of 0.933. The contributions for steric, electrostatic, and hydrophobic fields were 25, 44, and 31%. As a result, it was proved that the basic S4-directed substituents should be replaced against more hydrophobic building blocks to improve pharmacokinetic properties. The structural and chemical interpretation of CoMFA and CoMSIA contour maps directly pointed to those regions in the Factor Xa binding site, where steric, electronic, or hydrophobic effects play a dominant role in ligand-receptor interactions. 

Our studies also suggested that H-bond acceptors play an important role for compounds that bind the HERG channel (GRIND descriptors 13 and 34 in both the pharmacophoric models, as shown in Fig. 9.8). The statistical relevance of the MIFs generated by the hydrophobic probe confirmed the assumptions that were made in a previous CoMSiA model [17] regarding the presence of a hydrophobic feature. 

Fig. 6 Correlation plot for CoMFA and CoMSIA models with experimental enzyme inhibition data on human KMO enzyme...

Statistical characteristics of CMF models obtained for these data sets were compared with the same characteristics built for corresponding data sets using the common 3D-QSAR methods, CoMFA (Comparative Molecular Fields Analysis) [18] and CoMSIA (Comparative Molecular Similarity Index Analysis) [24], based on the use of molecular fields. Data on CoMFA and CoMSIA models were taken from Ref. [25]. 

Statistical parameters of CMF, CoMFA and CoMSIA models are shown in Table 13.3. They include the values of 4 statistical parameters and RMSE characterizing internal predictive performance, and RMSE —external predictive... 

CoMSIA (in Ref. [25]—CoMSIA2) models are based on electrostatic and steric fields molecular fields and also involve contributions from the hydrophobic and two hydrogen-bonding molecular fields. All CMF models are based on the use of all afore-mentioned five types of molecular fields. All CoMFA and CoMSIA models were obtained by using a lattice with 2 A spacing expanding at least 4 A in each direction beyond aligned molecules. Only the most predictive CoMFA and CoMSIA models are included in the Table 13.3. 

As it is clear from Table 13.3, models built for 7 data sets by using the CMF approach almost in all cases show better internal (cross-vahdation) predictive performance (i.e., higher q ) then the corresponding models obtained by the (IbMFAand CoMSIA methods. The values of the CMF models obtained for the COX-2 and BZR data sets are, respectively, equal and lower as compared to the corresponding CoMSIA models. 

There is also a moderate advantage in external predictive performance (estimated on external test sets using the parameters and RMSE of the CMF models over the CoMFA, models for 5 data sets (ACE, AChE, BZR, DHFR, and GPB), and CoMSIA models for 4 data sets (ACE, AChE, BZR and DHFR). 

As follows from Table 13.4, in all cases the value which characterizes the external predictive performance, is lower than the value q computed using the internal cross-validation. This means that the use of only two adjustable hyperparameters may cause the model selection bias . Almost in all cases the value q lies between and q. It is interesting to note that q for CMF models are usually higher than for CoMFA and CoMSIA models. The predictive performance assessed using the external 5-fold cross-validation procedure is especially high for ACE, DHFR and THR. 

Eleven compounds that were not included in the training set were selected as a test data set to validate the QSAR models. All of the test compounds were well predicted. The mean and standard deviation of prediction errors were 0.28 and 0.005 for the CoMFA model, and only 0.33 and 0.011 for the CoMSIA model. The predictive which was analogous to the cross-validated correlation coefficient q, was 0.883 for the CoMFA and 0.908 for the CoMSIA, suggesting a high reliability of these models. 

The consistency between the CoMFA/CoMSIA field distributions and the 3D topology of the protein stmcture indicated the robustness of the 3D-QSAR models. Overall, the degree of predictability of the CoMSIA model appeared to be similar to that of the CoMFA model. The combined use of both the CoMFA and the CoMSIA model might be more suitable for the prediction of the activities of novel designed compounds [22]. 

---