You are here:
PREDICTING TOXICOLOGICAL ENDPOINTS OF CHEMICALS USING QUANTITATIVE STRUCTURE-ACTIVITY RELATIONSHIPS (QSARS)
Citation:
HARTEN, P. F., D. M. YOUNG, T. M. MARTIN, AND R. VENKATAPATHY. PREDICTING TOXICOLOGICAL ENDPOINTS OF CHEMICALS USING QUANTITATIVE STRUCTURE-ACTIVITY RELATIONSHIPS (QSARS). Presented at EPA Science Forum 2006, Washington, DC, May 16 - 18, 2006.
Impact/Purpose:
To inform the public.
Description:
Quantitative structure-activity relationships (QSARs) are being developed to predict the toxicological endpoints for untested chemicals similar in structure to chemicals that have known experimental toxicological data. Based on a very large number of predetermined descriptors, an analysis finding chemical slusters of similar descriptors will be performed utilizing Ward's method. For all chemical slusters that have a reasonable number of chemicals, a QSAR will be developed to predict a toxicological endpoint for untested chemicals of similar structure. Optimal combinations of the descriptors will be found by implementing a genetic algorithm that searches through the space of all predetermined descriptors to find what combinations of these descriptors produce the most accurate QSAR prediction. Each QSAR developed will have to meet certain statistical criteria to be considered valid. Because the search through descriptor space involves millions upon millions of calculations, this research will effectively utilize EPA's high-performance computing resources.