Quantitative Characterization of Molecular Similarity Spaces: Tools for Computational Toxicology (1996-1997)

Abstract

Four classes of theoretical structural parameters, viz., topostructural, topochemical,geometrical and quantum chemical descriptors, have been used in the development of quantitative structure-activity relationship (QSAR) models for a set of sixty-nine benzene derivatives. None of the individual classes of parameters was very effective in predicting toxicity. A hierarchical approach was followed in using a combination of the four classes of indices in QSAR model development. The results show that the hierarchical QSAR approach using the algorithmically derived molecular descriptors can estimate the LC50 values of the benzene derivatives reasonably well.