Abstract
More realistic description of ligand-receptor interactions in CoMFA results from alignments considering surface and field properties instead of only molecular frameworks. The field fit algorithm implemented in SYBYL (Tripos Ass.) as part of the energy minimizer provides the possibility to assign individual weights to grid points. A new weighting function derives the significance of grid points ...
Abstract
More realistic description of ligand-receptor interactions in CoMFA results from alignments considering surface and field properties instead of only molecular frameworks. The field fit algorithm implemented in SYBYL (Tripos Ass.) as part of the energy minimizer provides the possibility to assign individual weights to grid points. A new weighting function derives the significance of grid points for the alignment of fields from preceding CoMFA runs, using regression coefficients, means, and standard deviations of field variables as parameters. Just in strongly diverse congeneric series, the method does not underestimate the common structure and not overweight variable, interacting regions. CoMFA of a large series of 142 histamine H2 receptor agonistic imidazolylpropylguanidines (pD2 values from guinea pig atrium) is presented as example. Results with three different alignments were compared: (1) exact superposition of the constant imidazolylpropylguanidine moiety, (2) SUPERIMPOSE or FIT of energy minima, (3) minimization of the structures by weighted field fit with weights based on CoMFA with alignment 2. A significant improvement of cross-validated PLS results was observed from alignment to alignment: Leave-one-out approach: (1) 7 PC's, Q2=0.59, sPRESS=0.50, (2) 8 PC's, Q2=0.66, sPRESS=0.46, (3) 9 PC's, Q2=0.71, sPRESS=0.42. Cross validation with 10 groups (mean of 10 runs): (1) 6.3 PC's, Q2=0.59, sPRESS=0.50, (2) 6.1 PC's, Q2=0.65, sPRESS=0.47, (3) 9.5 PC's, Q2=0.71, sPRESS=0.43. It is concluded that risks of the field fit method like producing artificial redundancy of the structures and ignoring entropy contributions to the free energy of binding are lowered with the given weighting method.
More literature (via CORE)