Appendix F – Software tool for toxicity prediction of pesticides, candidate pesticides, and their derivatives (user guide)

The aim of this guide is the description of a software tool for toxicity prediction of molecules of pesticides and related compounds. The software is based on the integration of the knowledge acquired in an EU project in a homogeneous manner using the best algorithms obtained as the basis for hybrid combinative models to be used for predictive purposes. The tool allows processing of chemical compounds one by one and in batch for the prediction of toxicity against five endpoints: acute toxicity for rainbow trout (Oncorhynchus mykiss): LC50 96-h exposure; acute toxicity for water flea (Daphnia magna): LC50 48-h exposure; acute oral toxicity for bobwhite quail (Colinus virginianus): LD50 14-day exposure; dietary toxicity for bobwhite quail (C. virginianus): LD50 8-day exposure; acute contact toxicity for honey bee (Apis melifera): LD50 48-h exposure. The input is the chemical structure of the molecule, characterized by a list of numerical chemical descriptors. The algorithms, as quantitative structure activity relationships (QSARs), output the estimated toxicity value. The values outside the expertise domain of the predictive models are also indicated.