The Prediction of the Isoelectric Point Value of Peptides and Proteins with a Wide Range of Chemical Modifications

Abstract

The scale of virtual pKa values for calculating the isoelectric point of peptides and proteins with chemical and post-translational modifications (PTM) is presented. The learning set of pKa values is based on data from 25 experiments of isoelectric focusing of peptides with subsequent mass spectrometric identification (ProteomeXchange accession codes: PXD000065, PXD005410, PXD006291, PXD010006 and PXD017201). In order to enrich the resulting sets with peptides containing modifications the identification of peptides was repeated using raw mass spectrometry data of all datasets. In the final learning set have included peptides satisfying the following conditions: the peptide was found in the fraction with scoring function maximum and maximum peptide abundance; the peptide was found in more than one experiment, and differences of the pI value between experiments was less than 0.15 pH unit. Two variants of the scales were created. In the first variant, pKa values depended only on the residue position relative to the ends of the sequence (N- or C-terminal residue or inside the chain). In the second variant, the effect of neighboring residues was also taken into account. The prediction accuracy of the second variant was higher. The comparison with other methods of pI prediction was carried out. Although the scale was calculated from set containing only peptides, it would be applicable for pI prediction of proteins with and without PTM. The software for prediction of pI values using the resulting pKa scales is available at http://pIPredict3.ibmc.msk.ru.