Linear solvation energy relationships as classifier in non-target analysis--an approach for isocratic liquid...

Linear solvation energy relationship (LSER) models are developed to characterize various HPLC columns and to predict retention of candidate compounds in non-target analysis of environmental samples. The models can be applied as classifier to exclude potential candidate structures with the same empirical formula. The application of the classifier was tested for a set of 19 isomers with empirical formula C12H10O2. The performance of the classifier was also demonstrated for a validation set of seven structural diverse compounds, where all potential candidates for the corresponding empirical formulas were included in the studies. Herein, the standard deviation of all compound descriptors of the respective candidate compounds was used as indication for potential exclusions on a certain column.