In Silico Prediction of Physicochemical Properties

This report provides a critical review of computational models, and in particular(quantitative) structure-property relationship (QSPR) models, that are available for the prediction of physicochemical properties. The emphasis of the review is on the usefulness of the models for the regulatory assessment of chemicals, particularly for the purposes of the new European legislation for the Registration, Evaluation, Authorisation and Restriction of CHemicals (REACH), which entered into force in the European Union (EU) on 1 June 2007. It is estimated that some 30,000 chemicals will need to be further assessed under REACH. Clearly, the cost of determining the toxicological and ecotoxicological effects, the distribution and fate of 30,000 chemicals would be enormous. However, the legislation makes it clear that testing need not be carried out if adequate data can be obtained through information exchange between manufacturers, from in vitro testing, and from in silico predictions. The effects of a chemical on a living organism or on its distribution in the environment is controlled by the physicochemical properties of the chemical. Important physicochemical properties in this respect are, for example, partition coefficient, aqueous solubility, vapour pressure and dissociation constant. Whilst all of these properties can be measured, it is much quicker and cheaper, and in many cases just as accurate, to calculate them by using dedicated software packages or by using (QSPRs). These in silico approaches are critically reviewed in this report.JRC.I.3-Toxicology and chemical substance