Probabilistic enrichment of phosphopeptides by their mass defect.

The mass defect, that is, the difference between the nominal and actual monoisotopic masses, of a phosphorus in a phosphate group is greater than for most other atoms present in proteins. When the mass defects of tryptic peptides derived from the human proteome are plotted against their masses, phosphopeptides tend to fall off the regression line. By calculating the masses of all potential tryptic peptides from the human proteome, we show that regions of higher phosphorylation probability exist on such a plot. We developed a transformation function to estimate the mass defect of a peptide from its monoisotopic mass and empirically defined a simple formula for a user-selectable discriminant line that categorizes a peptide mass according to its probability of being phosphorylated. Our method performs similarly well on phosphopeptides derived from a database of experimentally validated phosphoproteins. The method is relatively insensitive to mass measurement error of up to 20 ppm. The approach can be used with a tandem mass spectrometer in real time to rapidly select and rank order the possible phosphopeptides from a mixture of unmodified peptides for subsequent phosphorylation site mapping and peptide sequence analysis.