With the aim of applying recent quantitative structure-activity relationship (QSAR) descriptors coming from the pharmaceutical sciences to the modelization of ecotoxicity data, this work deals with the establishment of 3D-QSAR on chlorophenols by using for the first time in this field the recent CATALYST software. The training set has been intentionally chosen simply because of its homogeneous character and the abundance of reliable experimental values. Among 69 data sets collected in the literature on these chemicals, 9 were selected to have fulfilled CATALYST simulation conditions. The principle of CATALYST is briefly described. The different investigations have led to nine 3D-QSAR called "hypotheses," meeting the criteria of high statistical significance attested by costs and correlation coefficient values. The ecotoxicity values calculated from the 9 models established in this study. Analysis of the graphical representations of hypotheses led to hypotheses about the toxicity mechanisms. A comparison is finally made between the CATALYST models and those previously derived by other modeling methods.
Copyright 1999 Academic Press.