Documentation scienceplus.abes.fr version Bêta
AttributsValeurs
type
Is Part Of
Subject
Title
  • Reliability of Comparative Molecular Field Analysis Models: Effects of DataScaling and Variable Selection Using a Set of Human Synovial FluidPhospholipase A2 Inhibitors
has manifestation of work
related by
Author
Abstract
  • The effects of data pretreatment, data scaling, and variableselection on three-dimensionalquantitative structure−activity relationships derived by comparativemolecular field analysis(CoMFA) using the GRID energy function were studied in detail for a setof inhibitors of thehuman synovial fluid phospholipase A2(HSF-PLA2). The quality of the models wasevaluatedfor predictive power and ability to map the receptor binding site by(a) comparison of predictedand experimental activities using cross-validation and externalvalidation sets and (b)comparison of the regions selected in space in the CoMFA models with acrystal structure ofa HSF-PLA2−inhibitor complex, with optimized comparativebinding energy analysis (COMBINE) models (Ortiz et al., 1995) and withstructure−activity relationships derived previouslyfor different sets of compounds. It is found that (1) data scalingand dielectric modeling stronglyinfluence CoMFA results. Unscaled data and a uniform dielectricconstant of 4 are well suitedto GRID-CoMFA studies for the present compound set. (2) The GOLPEand Q2-GRS variableselection methods select variables in roughly the same regions inCartesian space, but theyproduce different models in chemometric space and differ in theirsensitivity to data scalingand pretreatment and their tendency to overfitting. (3) CoMFAmodels are consistent withCOMBINE models in that they identify approximately the sameintermolecular interactionsas relevant for activity. Our study provides support for thequalitative receptor-mappingproperties of CoMFA models and for the validity of variable selectionwhen applied with careand also provides guidelines for how to evaluate the quality of CoMFAmodels.
article type
is part of this journal



Alternative Linked Data Documents: ODE     Content Formats:       RDF       ODATA       Microdata