Evaluation of data analysis strategies for improved mass spectrometry-based phosphoproteomics

Here we describe a set of enhanced data processing and filtering methods to improve significance and coverage of phosphopeptide identifications by mass spectrometry. We demonstrate that for samples of limited complexity, spectra-based estimation of false discovery rates will lead to overprediction of confidently identified phosphorylated peptides due to a bias caused by multiple fragmentation of highly abundant peptide species. We further provide evidence that fragmentation of abundant peptides at the tails of their chromatographic peaks is a major source for false positive peptide matches and that overall confidence in phosphopeptide identifications can be improved by a chromatographic peak-based aggregation scheme, intensity rank-based neutral loss and optimized mass error filters. When replicate runs of a standard sample were performed using different fragmentation techniques on an Orbitrap mass spectrometer we observed improvements of 7-31% in phosphopeptide coverage depending on the fragmentation method and the desired false discovery rate.