Quantitative structure - Activity relationship studies of progesterone receptor binding steroids

The selection of appropriate descriptors is an important step in the succes sful formulation of quantitative structure-activity relationships (QSARs). This paper compares a number of feature selection routines and mapping meth ods that are in current use. They include forward stepping regression (FSR) , genetic function approximation (GFA), generalized simulated annealing (GS A), and genetic neural network (GNN). On the basis of a data set of steroid s of known in vitro binding affinity to the progsterone receptor, a number of QSAR models are constructed. A comparison of the predictive qualities fo r both training and test compounds demonstrates that the GNN protocol achie ves the best results among the 2D QSAR that are considered. Analysis of the choice of descriptors by the GNN method shows that the results are consist ent with established SARs on this series of compounds.