Prediction of liquid chromatographic retention behavior based on quantum chemical parameters using supervised self organizing maps

Abstract

Self organizing maps (SOMs) in a supervised mode were applied for prediction of liquid chromatographic retention behavior of chemical compounds based on their quantum chemical information. The proposed algorithm was simple and required only a small alteration of the standard SOM algorithm. The application was illustrated by the prediction of the retention indices of bifunctionally substituted N-benzylideneanilines (NBA) and the prediction of the retention factors of some pesticides. Although the predictive ability of the supervised SOM could not be significantly greater than that of some previously established neural network methods, such as a radial basis function (RBF) neural network and a back-propagation artificial neural network (ANN), the main advantage of the proposed method was the ability to reveal non-linear structure of the model. The complex relationships between samples could be visualized using U-matrix and the influence of each variable on the predictive model could be investigated using component planes—which can provide chemical insight.