Ecotoxicty

The risk assessment of pesticides / chemicals in toxicology and environmental chemistry remains a permanent challenge. In fact, over the last fifty years, a large amount of bioactive compounds have been designed and synthesized. Systematic experimental determination of toxicity would require incredibly high resources. Predictive tools based on computational approaches represent therefore a valid alternative, as recognized by many international regulatory bodies.

We have successfully applied our proprietary data mining methods, inside European Commission long term contracts, on data sets ranging from 200 to 600 chemicals or pesticides, active against several end-points, such as rat, trout, fathead minnow. After establishing a rigorous procedure of model validation, we achieved up to 80% in prediction accuracy of toxicity range for the test set compounds.