Prediction of the fish acute toxicity from heterogeneous data coming from notification files.

Four descriptors (Molecular weight, log(Pow), hardness and free energy of s olvation) were selected to predict, on a training set of heterogeneous chem ical compounds, the fish acute toxicity. The data were extracted from 523 n otification files of new chemicals stored at the French Department of the E nvironment. The selection of the descriptors was carried out by using a sta tistical technique coupling OLS regression and genetic algorithm. The limit s of validity for the final equation are discussed by comparing the actual and predicted activities on several compounds. (C) 1999 Elsevier Science Lt d. Ail rights reserved.