|
Internet Electronic Journal of Molecular Design - IEJMD, ISSN 1538-6414, CODEN IEJMAT
| ABSTRACT - Internet Electron. J. Mol. Des. October 2002, Volume 1, Number 10, 559-571 |
QSAR Models for the Dermal Penetration of Polycyclic Aromatic Hydrocarbons
Ovidiu Ivanciuc, Teodora Ivanciuc, and Alexandru T. Balaban
Internet Electron. J. Mol. Des. 2002, 1, 559-571
|
Abstract:
Quantitative structure-activity relationships (QSAR) models for the
dermal penetration of 60 polycyclic aromatic hydrocarbons
(PAHs) were established with the CODESSA program. The QSAR
models for the PAH dermal penetration are obtained by selecting
descriptors from a wide diversity of constitutional, topological,
electrostatic and quantum structural indices. Standard quantum
chemistry packages are used for optimizing the molecular
geometry and for semi-empirical quantum computations. A
heuristic algorithm selects the best multiple linear regression
equation according to the highest statistical indices; the predictive
power of each QSAR is estimated with the leave-one-out (LOO)
cross-validation method. The best QSAR model with two
descriptors has r2 = 0.748, r2LOO
= 0.719, s = 6.5, and F = 84.42. A
significant improvement of the statistical indices is obtained by
adding a third theoretical descriptor, i.e. r2
= 0.761, r2LOO = 0.725,
s = 6.4, and F = 59.39. Our results demonstrate that QSAR models
can be used in risk assessment studies in order to estimate the
dermal penetration properties of PAHs from mineral oils, coal, tar,
and derived products.
|