QUANTITATIVE RELATIONSHIP BETWEEN THE PHYSIOCHEMICAL CHARACTERISTICS AS WELL AS GENOTOXICITY OF ORGANIC POLLUTANTS AND MOLECULAR AUTOCORRELATION TOPOLOGICAL DESCRIPTORS

Autocorrelation topological indices were used as molecular descriptors to study the relationship between physiochemical characteristics as w ell as genotoxicity of organic pollutants and the molecular structures . Significant quantitative structure-activity relationships were estab lished Introduction Quantitative structure-activity relationship (QSAR ) studies are useful in assessing environmental effects of chemicals, and different mathematical models have been proposed in order to evalu ate the link between structures and characteristics of organic polluta nts.[1] Various molecular topological descriptors have been widely app lied in describing molecular structure. Among them, the molecular conn ectivity indices, proposed by Kier, are mainly used in the environment al field; while, the autocorrelation topological indices, based on the autocorrelation mathematic function, are often used in a pharmacologi cal SAR study, but play few roles in environmental science[2-5]. Becau se molecular autocorrelation topological indices do not requre any res trictive hypothesis on the nature of molecules and can be manipulated conveniently, it appears very promising in physiochemical parameters a nd ecotoxicological QSAR investigations. In the article, the calculati on of autocorrelation topological index was revised slightly based on the conventional method, and the prediction ability was enhanced signi ficantly. The relationship was well established between solubilities i n water (Sw) as well as Octanol/Water partition coefficients (Kow) of 55 substituted benzenes and autocorrelation topological indexes. In ad dition, the genotoxicity of 26 hydrocarbons was studied through micron uclei test, with 20 compounds active, and their MNI (20 parts per thou sand) were deduced. A satisfying regression equation was obtained betw een the MNI and autocorrelation topological index. (C) 1997 Elsevier S cience Ltd.