|
Internet Electronic Journal of Molecular Design - IEJMD, ISSN 1538-6414, CODEN IEJMAT
| ABSTRACT - Internet Electron. J. Mol. Des. March 2002, Volume 1, Number 3, 157-172 |
Support Vector Machine Identification of the Aquatic Toxicity Mechanism
of Organic Compounds
Ovidiu Ivanciuc
Internet Electron. J. Mol. Des. 2002, 1, 157-172
|
Abstract:
Because numerous organic chemicals can be environmental
pollutants, considerable efforts were directed towards the
study of the relationships between a compound's structure
and its toxicity. Significant progress has been made to
classify chemical compounds according to their mechanism
of toxicity and to screen them for their environmental risk
assessment. The prediction of the mechanism of action using
structural descriptors has major applications in selecting the
appropriate quantitative structure-activity relationships
(QSAR) model, to identify chemicals with similar toxicity
mechanism, and in extrapolating toxic effects between
different species and exposure regimes. Support vector
machine (SVM) is a new machine learning algorithm that
found numerous applications in various classification studies.
In this study we have investigated the application of SVM for
the recognition of the aquatic toxicity mechanism of 88
organic compounds. For each compound, the chemical
structure was encoded by four structural descriptors, namely
the octanol-water partition coefficient log Kow, the energy of
the highest occupied molecular orbital EHOMO, the energy of
the highest unoccupied molecular orbital ELUMO, and the
average acceptor superdelocalizability SNav. Extensive
simulations using the dot, polynomial, radial basis function,
neural, and anova kernels demonstrate that the classification
performances of SVM depend strongly on the kernel type and
various parameters that control the kernel shape. The best
prediction results were obtained with a polynomial kernel of
degree 2. Support vector machines represent a powerful and
flexible classification algorithm, with many potential
applications in QSAR and molecular design. The results
reported in the present study demonstrate such an application
in the identification of the aquatic toxicity mechanism.
|