Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Development of chemical categories by optimized clustering strategies

 
: Bitsch, Annette; Batke, Monika; Gundert-Remy, U.; Gütlein, M.; Kramer, S.; Partosch, F.; Seeland, M.

The Toxicologist 54 (2015), No.1, pp.480, Abstract PS 2236
ISSN: 0731-9193
Society of Toxicology (Annual Meeting) <54, 2015, San Diego/Calif.>
English
Abstract
Fraunhofer ITEM ()

Abstract
According to OECD a chemical category is a group of chemicals whose physicochemical and human health and/or ecotoxicological properties are likely to be similar or follow a regular pattern. The building of categories has often been tried on the basis of conventional structure based approaches. In the present project we developed an approach by which toxicological and structural properties likewise contribute to the building of chemical categories for (sub)chronic toxicity. Two databases on repeated-dose toxicity (RepDose and the "ELINCS" data base) served as data basis. The toxicological data are organized into organ toxicity split into subgroups according to phenotypic and mechanistic observations. For the definition of a category, the following characteristics were considered: organ investigated, effects, no effects; potency in terms of no observed adverse effect level (NOAEL), organ specificity. A multi-label clustering by using predictive clustering trees (PCT) was established. Several decisions concerning structural features and chemicals properties as well as the toxicological data had to be considered during development: - the selection of features and their SMARTS description - the non-use of PC parameters - imputation methods for missing values - the level of detail for a consistency of toxicological data versus data density in the matrix. All resulting category clusters were visualized and checked for plausibility. An important decision about a stop criterion for clustering was the use of toxicological variance data in combination with statistical significance. In the process of developing this approach we needed many incremental improvements; the final approach shows a set of useful and representative clusters now. This project was funded by BMBF.

: http://publica.fraunhofer.de/documents/N-349462.html