|
Internet Electronic Journal of Molecular Design - IEJMD, ISSN 1538-6414, CODEN IEJMAT
| ABSTRACT - Internet Electron. J. Mol. Des. November 2002, Volume 1, Number 11, 610-619 |
A Novel Variable Selection and Modeling Method based on the
Prediction for QSAR of Cyclooxygenase-2 Inhibition by Thiazolone
and Oxazolone Series
Shu-Shen Liu, Shi-Hai Cui, Yun-Yu Shi, and Lian-Sheng Wang
Internet Electron. J. Mol. Des. 2002, 1, 610-619
|
Abstract:
A novel variable selection and modeling method based on prediction is
developed to construct the quantitative structure-activity relationships
(QSAR) between the molecular electronegativity distance vector based
on 13 atomic types and the biological activities of a set of selective
cyclooxygenase-2 inhibitory molecules, thiazolone and oxazolone
series (TOS). Two parameters, the interrelation coefficient between the
independent variables and the correlation coefficient in the
leave-one-out cross-validation (q), are introduced into the classical
all-subset regression to improve its searching course. Using multiple linear
regression, a 4-variable linear model for the data set of 21 TOSs is
developed with the correlation coefficient of 0.9248 and the root mean
square error of 0.283 in modeling stage and the coefficient of 0.8849
and the error of 0.351 in the leave-one-out validation step,
respectively. To further test the predictive ability of the model, 15
TOS compounds are picked up from the set of 21 TOSs to construct a
training set which is used to build a QSAR model and then the model
is employed to predict the biological activities of the remaining
compounds.
|