A CoMFA analysis with conformational propensity: An attempt to analyze theSAR of a set of molecules with different conformational flexibility using a 3D-QSAR method
K. Gohda et al., A CoMFA analysis with conformational propensity: An attempt to analyze theSAR of a set of molecules with different conformational flexibility using a 3D-QSAR method, J COMPUT A, 14(3), 2000, pp. 265-275
CoMFA analysis, a widely used 3D-QSAR method, has limitations to handle a s
et of SAR data containing diverse conformational flexibility since it does
not explicitly include the conformational entropic effects into the analysi
s. Here, we present an attempt to incorporate the conformational entropy ef
fects of a molecule into a 3D-QSAR analysis. Our attempt is based on the as
sumption that the conformational entropic loss of a ligand upon making a li
gand-receptor complex is small if the ligand in an unbound state has a conf
ormational propensity to adopt an active conformation in a complex state. F
or a QSAR analysis, this assumption was interpreted as follows: a potent li
gand should have a higher conformational propensity to adopt an `active-con
formation'-like structure in an unbound state than an inactive one. The con
formational propensity value was defined as the populational ratio, N-activ
e/N-stable, of the number of energetically stable conformers, N-stable, to
the number of 'active-conformation'-like structures, N-active. The latter n
umber was calculated by counting the number of conformers that satisfied th
e structural parameters deduced from the active conformation. A set of SAR
data of imidazoleglycerol phosphate dehydratase inhibitors containing 20 mo
lecules with different conformational flexibility was used as a training se
t for developing a 3D structure-activity relationship by a CoMFA analysis w
ith the conformational propensity value. This resulted in a cross-validated
squared correlation coefficient of the CoMFA model with the conformational
propensity value (R = 0.640) higher than that of the standard CoMFA model
(R = 0.431). Then we evaluated the quality of the CoMFA models by predictin
g the inhibitory activity for a new molecule.