2D Autocorrelation modeling of the activity of trihalobenzocycloheptapyridine analogues as farnesyl protein transferase inhibitors

The inhibitory activity towards farnesyl protein transferase enzyme ( FPT) of 49 piperidine substituted trihalobenzocycloheptapyridine analogues ( thBCHPs) has been successfully modelled using 2D spatial autocorrelation vectors. Predictive linear and non- linear models were obtained by forward stepwise multilinear regression analysis ( MRA) and artificial neural network ( ANN) approaches, respectively. A variable selection routine that selected relevant non- linear information from the data set was employed prior to networks training. The MRA model, using three descriptors, was able to explain about 68% data variance. The model showed a linear dependence between the inhibitory activities and autocorrelation coefficients weighted by van derWaals volumes and atomic polarizabilities on the inhibitors molecules. The non- linear approach preserve several characteristics described for the linear one. Three descriptors were selected encoding the same atomic properties, but the new ones were able to explain about 92% data variance. In addition, the ANN model had higher predictive power. Furthermore, inhibitors were well distributed regarding its activity levels in a Kohonen self- organizing map ( SOM) built using the input variables of the best neural network.