You are here:
ITERATIVE PROCESS OF QSAR BUILDING AND STRATEGIC TESTING: PREDICTING ER BINDING AFFINITY
Citation:
Nedyalkova, Z., M. Todorov, H. Aladjov, J. Atanasov, P. K. Schmieder, G D. Veith, AND O. G. Mekenyan. ITERATIVE PROCESS OF QSAR BUILDING AND STRATEGIC TESTING: PREDICTING ER BINDING AFFINITY. Presented at SETAC Europe, Hamburg, Germany, April 27-May 1, 2003.
Description:
Basic principles of QSAR model development and application are discussed. The most difficult step in QSAR application for regulatory use may be determining when a model is sufficiently improved to provide predictions for a specified chemical domain of regulatory concern. The iterative scheme presented for QSAR model development and appropriate refinement includes the following steps: model building, assessment, strategic testing, and model improvement to meet acceptance criteria. Tools are being developed to apply this scheme to predict chemical binding to the estrogen receptor - - - The application of QSAR in strategic testing to increase domain coverage will also be discussed, including the following options: 1) selecting chemicals from the farthest reaches of structure space with respect to the TrSet; 2) evenly selecting chemicals throughout the model extrapolation domain: or, 3) choosing chemicals with a probability to be active above a user-defined threshold and falling into the TrSet coverage domain with a given probability.