| |
[摘要]：A 'sum-model' (3D QSAR - CoMFA) has been developed to design PPAR(alpha/gamma/delta) (peroxisome proliferator activated receptor) pan agonists by using the sum of activities (EC50) of compounds against individual subtypes as a dependent parameter. In addition, the three subtype specific CoMFA models were also generated using the identical training set molecules (N = 28). All four models were validated using the popular 'leave-one-out' (LOO) method and with a test set of 9 molecules. The generated models were found to be statistically significant with r(cv)(2) > 0.5 and r(ncv)(2) > 0.9 and the lower values of standard error of estimation (SEE) ranging from 0.097 to 0.160. From the contour map analyses the 'sum model' was found to represent the three subtype specific models and also predicted the sum of activities of the training set molecules with reasonable accuracy. The new molecules were designed based on the 'sum-model' and were found to dock well in the PPAR gamma active site. This approach may find wider applications in the research related to other classes of 'designed multiple ligands'. (C) 2008 Elsevier Masson SAS. All rights reserved. |
|
