Prediction of peptide retention time in reversed-phase high-performance liquid chromatography

Peptide retention in reversed-phase chromatography depends mainly on the amino acid composition of peptides and can therefore be predicted by summing the relative hydrophobic contributions of each constitutive amino acid residue. The prediction is correct for small peptides but overestimates the retention times of peptides larger than 10-15 residues. A new prediction model is proposed in which the contribution to peptide retention of each amino acid residue is not a constant but a decreasing function of peptide length. From the retention times of 104 peptides, the parameters of decreasing functions were estimated by a non-linear multiple regression analysis. The contribution to peptide retention of charged, polar and non-polar residues appears to be differently affected by peptide length. The secondary structure of most peptides during reversed-phase high-performance liquid chromatography could be responsible for this. The high correlation between the predicted and observed retention times of peptides which were not used to establish the model indicates a good predictive accuracy of the new model.