engleski

DEVELOPMENT OF GRADIENT RETENTION MODEL IN ION CHROMATOGRAPHY: APPLICATION OF ANFIS DEVELOPED QSRR MODELS

Application of quantitative structure-retention relationships (QSRR) models, which predicts chromatographic retention explicitly on basis of analytes’ molecular structure, can generally short the time needed for selection of appropriate chromatographic method when dealing with other similar analytes. Also it can replace time-consuming “trial and error” approach in method optimization. Development of gradient retention model in ion chromatography using QSRR methodology gathered with adaptive neuro fuzzy inference system (ANFIS) modeling tool was presented in this work. The used analytes were carbohydrates. A large set of physico-chemical variables, called descriptors, was determined from the analytes’ molecular structure. Among them, descriptors with the highest content of useful information were selected by applying several elimination criteria and thereafter genetic algorithm optimization tool. ANFIS was applied in order to develop good and reliable QSRR model. The QSRR models were developed for four isocratic elutions and thereafter applied for retention prediction of other similar analytes, i.e. test components. The predicted retention values were used in obtaining the general isocratic model for each of the test components. In order to enable prediction for gradient elutions, a gradient model based on isocratic data was applied.

GRADIENT RETENTION MODEL; ION CHROMATOGRAPHY; ANFIS; QSRR MODELS

nije evidentirano