Computational toxicology using OpenTox & Bioclipse

Abstract

Computational methods, e.g., OECD QSAR Toolbox and ToxPredict, are increasingly used to support chemicals toxicity assessment in academic settings, industry and governments. The in silico-based assessments complement experimental approaches and have potential to fill the knowledge gaps needed to broadly assess chemical hazards. We present here the interoperable Bioclipse–OpenTox platform as a novel alternative made freely and openly available.The interactive Bioclipse software is combined with remote computational toxicity prediction provided by services in the OpenTox network from various European institutes and SMEs. These online services apply machine learning methods for integration of a number of end points, e.g., Ames mutagenicity test in salmonella, Caco-2 cell model permeability and micronucleus assay in rodents. Coupling of such data to chemical structure assessments lead to prediction of site(s) for metabolism (using SMARTCyp), biodegradation (START), and toxicity mode prediction (Verhaar scheme). The OpenTox platform thus unifies how the services are accessed whereas the Bioclipse software provides the easy-to-use interface for interactively studying the toxic part of molecules. Additional predictive methods that are made accessible via the OpenTox network are automatically discovered by Bioclipse and models can be improved over time without any need to reinstall Bioclipse itself.