QSAR modeling of AT1 receptor antagonists using ANN.

Multiple linear regression (MLR) and artificial neural networks (ANN) have been used for structure-activity relationship analysis for a set of 113 AT1 receptor antagonists. The ANN model showed better performance with a 6-6-1 architecture than MLR. The results obtained from this study indicate that three descriptors, hydration energy (EH), n-octanol/water partition (LOGP), and energy of the lowest unoccupied molecular orbital (LUMO), play an important role on the activity of AT1 receptor antagonists with biphenyltetrazole structures. This information is pertinent to the further design of new AT1 receptor antagonists. 1 A plot of observed versus predicted PIC50 produced from the best nonlinear model using 6-6-1 architecture.