Review on modelling aspects in reversed-phase liquid chromatographic quantitative structure–retention relationships

In the literature an increasing interest in quantitative structure–retention relationships (QSRR) can be observed. After a short introduction on QSRR and other strategies proposed to deal with the starting point selection problem prior to method development in reversed-phase liquid chromatography, a number of interesting papers is reviewed, dealing with QSRR models for reversed-phase liquid chromatography. The main focus in this review paper is put on the different modelling methodologies applied and the molecular descriptors used in the QSRR approaches. Besides two semi-quantitative approaches (i.e. principal component analysis, and decision trees), these methodologies include artificial neural networks, partial least squares, uninformative variable elimination partial least squares, stochastic gradient boosting for tree-based models, random forests, genetic algorithms, multivariate adaptive regression splines, and two-step multivariate adaptive regression splines.