You are here:
IN SILICO METHODOLOGIES FOR PREDICTIVE EVALUATION OF TOXICITY BASED ON INTEGRATION OF DATABASES
Citation:
Yang, C. AND A M. Richard. IN SILICO METHODOLOGIES FOR PREDICTIVE EVALUATION OF TOXICITY BASED ON INTEGRATION OF DATABASES. Presented at American Chemical Society, New Orleans, LA, March 23-27, 2003.
Description:
In silico methodologies for predictive evaluation of toxicity based on integration of databases
Chihae Yang1 and Ann M. Richard2, 1LeadScope, Inc. 1245 Kinnear Rd. Columbus, OH. 43212 2National Health & Environmental Effects Research Lab, U.S. EPA, Research Triangle Park, NC 27711.
The ability to accurately "predict" toxicity with in silico methods is increasingly emphasized as industry moves toward efficient up-front screening to reduce late stage attrition in drug discovery. However, current methods for structure-based toxicity estimation are not yet satisfactorily predictive. Reasons include the intrinsically complex nature of chemically induced toxicity and the lack of data from which the "models" or "predictions" are derived. Large amounts of toxicity information are publicly available; however, most of these databases are not optimized for building structure-toxicity relationships, let alone QSAR. The relationship between quality of data and prediction model accuracy intensifies the need for improved access to quality toxicity information. This paper describes collaboration between an EPA-sponsored public initiative, DSSTox (Distributed Structure Searchable Toxicity) database network, and a private sector effort, LIST (LeadScope In Silico Tox) focus group. Both are working towards improved data access and the integration of disparate data formats from various data sources. DSSTox is promoting SDF format for toxicity databases inclusive of chemical structures, whereas LIST is developing controlled toxicity vocabularies and mapping the data fields of SDF and XML schema. A sample carcinogenicity database has been built and used to derive models for predicting rodent carcinogenicity. Improving prediction capability by integration of data to enhance chemical space are shared objectives of the DSSTox and LIST initiatives. This abstract does not reflect EPA policy nor does mention of trade names indicate EPA endorsement.