|
Internet Electronic Journal of Molecular Design - IEJMD, ISSN 1538-6414, CODEN IEJMAT
| ABSTRACT - Internet Electron. J. Mol. Des. July 2004, Volume 3, Number 7, 426-442 |
Similarity Matrices Quantitative Structure-Activity Relationships for
Anticonvulsant Phenylacetanilides
Ovidiu Ivanciuc
Internet Electron. J. Mol. Des. 2004, 3, 426-442
|
Abstract:
Molecular graph descriptors are used in developing structure-property
models, in drug design, virtual synthesis, similarity
and diversity assessment. We present a new application of
topological indices in computing similarity matrices that are
subsequently used to develop quantitative structure-property
relationship and quantitative structure-activity relationship
models. The molecular structure is described by similarity
matrices obtained from similarity indices calculations, when
each molecule is compared to every other from the data set.
Four similarity indices are introduced for the computation of
the molecular similarity from a set of topological indices that
numerically characterize the structure of chemical compounds.
Using the multilinear regression model, the significant columns
from the similarity matrices are selected as independent
variables in a structure-activity study of anticonvulsant
phenylacetanilides. The results obtained show that similarity
matrices derived from molecular graph descriptors can provide
the basis for the investigation of quantitative structure-activity
relationships.
|