MODELING OF GAS-CHROMATOGRAPHIC RETENTION INDEXES USING COUNTERPROPAGATION NEURAL NETWORKS

Unspecific fragmentation of organic substances in the ion source of MS detector hinders identification of organic substances in gas chromato graphic separation. In such instances theoretical prediction of the re tention indices could be a useful tool, A new method for theoretical p rediction of gas chromatographic retention indices is described. Artif icial neural networks were trained in counterpropagation mode to predi ct retention data. Extensive data sets of simple organic compounds wit h known retention indices taken from the Literature were serving for t raining and test sets. The structure of molecules was described with a 12-dimensional vector the components of which were topological and ch emical parameters. Various geometries of artificial neural networks we re tested and different divisions into training and testing sets tried . The ANN with the configuration of 15 x 15 neurons has been chosen fo r routine work. The average RMS value was 36.6 retention time units.