Support Vector Regression Based QSPR for the Prediction of Retention Time of Peptides in Reversed-Phase Liquid Chromatography

In the present study, Quantitative Structure-Property Relationship (QSPR) models were developed to investigate the retention times (t(R)) of various peptides in seven reversed-phase liquid chromatography systems using Partial Least Squares (PLS), Artificial Neural Network (ANN) and Support Vector Machine (SVM) techniques. Different types of molecular descriptors were calculated to represent the molecular structures of the various compounds studied. Important descriptors were selected by a Genetic Algorithm-Partial Least Square (GA-PLS) method. The four descriptors selected using GA-PLS were used as inputs for PLS, ANN and SVM to build models to predict the retention times. Our study reveals that the relation between the chemical properties and retention time is a nonlinear phenomenon and that the PLS method is not capable to properly model it. The results obtained demonstrate that, for all seven data sets, the t(R) values estimated by SVM were in good agreement with the experimental data, and the performances of the SVM models were comparable or superior to those of PLS and ANN.