STUDIES OF IMIDAZOLE AND PYRAZOLE PROTONATION USING ELECTROSTATICALLYTRAINED NEURAL NETWORKS

A backpropagation neural network was trained with the molecular electr ostatic potentials (MEPs) of a series of substituted imidazoles to pre dict their corresponding pK(a). Using MEPs determined with a variety o f semiempirical and ab initio methods, the predictive power of the tra ined network was found to be sensitive to the quality of the basis set . The network was also trained to predict the proton affinity (E(pa)) and pK(a), both individually and combined, for a series of pyrazole MN DO MEPs.