Proteome informatics research group (iPRG)_2012: a study on detecting modified peptides in a complex mixture

Abstract

The proteome informatics research group of the Association of Biomolecular Resource Facilities conducted a study to assess the community's ability to detect and characterize peptides bearing a range of biologically occurring post-translational modifications when present in a complex peptide background. A data set derived from a mixture of synthetic peptides with biologically occurring modifications combined with a yeast whole cell lysate as background was distributed to a large group of researchers and their results were collectively analyzed. The results from the twenty-four participants, who represented a broad spectrum of experience levels with this type of data analysis, produced several important observations. First, there is significantly more variability in the ability to assess whether a results is significant than there is to determine the correct answer. Second, labile post-translational modifications, particularly tyrosine sulfation, present a challenge for most researchers. Finally, for modification site localization there are many tools being employed, but researchers are currently unsure of the reliability of the results these programs are producing.

Fig. 1.

Summary of peptide identification results reported by participants. Submitted results were compared with consensus results. Participants were also asked to indicate whether a particular result was above or below their estimated 1% FDR threshold. The plot separates results based on whether assignment were deemed correct according to consensus results and whether a result was reported as significant by the participant: blue: consensus identification better than 1% FDR threshold; green: consensus identification worse than 1% FDR threshold; red: nonconsensus result better than 1% FDR threshold; yellow: no consensus result (too few participants reporting a result to reach a consensus) better than 1% FDR; gray; nonconsensus result worse than 1% FDR threshold. The five participants to the right of the dotted line reported only modified peptides. Participant 23117 merged in an additional 2257 spectra derived from repeat MS/MS of the same precursor (each merged set counts here as only one spectrum), whereas everyone else reported individual peptide spectrum matches.

Fig. 2.

Co-isolated precursor ions led to different peptide identifications for a single spectrum depending on the source of the pre-processed peak list used for searching. In the supplied de-isotoped peak lists the spectrum was indicated as having the precursor m/z 465.19 2+; whereas in nonde-isotoped peaklists the precursor was indicated as m/z 464.59 3+. A, In the MS1 scan that triggered the MS/MS spectrum the 3+ precursor is < 5% as abundant as the 2+ precursor. B, The resulting spectrum is able to match y2-y7 of the peptide SVSDY(Nitro)EGK for the 2+ precursor ion and C, y3-y8, as well as doubly charged ions of y10-y12, to the peptide LAAPENEKPAPVR for the 3+ precursor ion (the y axis has been magnified relative to the base peak to allow easier visualization of the peaks).

Fig. 3.

Heat map plot reporting which peptides were identified by each participant. Each row represents one of the 70 spiked in synthetic peptides, which are grouped by modification type, whereas each column is a participant. Modification site localization was ignored for this plot.

Fig. 4.

Comparison of fragmentation spectra of a tyrosine phosphorylated and tyrosine sulfated peptide. The peptide DISLSDYK was synthesized with either a phosphorylation or sulfation on its tyrosine residue. The phosphopeptide spectrum contains ions y2-y6 where the phosphate group is retained, whereas the equivalent ions in the sulfopeptide spectrum are all observed 80 Da lower in mass because of prompt loss of SO3.

Fig. 5.

Summary of PTM site localization results by participants. For modified peptides, participants were required to indicate on which residue they believed the modification was localized and whether they were confident of the site localization. The top plot reports the number of spectra from the synthetic modified peptides that were identified by each participant. Green: correct site localization and confident assignment; red: incorrect site localization and confident assignment; gray: peptide contained additional modifications for which site localization is unknown but site localization was reported as confident; black: site localizations in peptide were reported as not confident. The lower panel reports a false localization rate (FLR) for reported results. This was calculated by dividing the number of confident incorrect site localization spectra (red portion of upper plot) by the total number of spectra for which the participant reported they were confident of site localizations. Values under the axis were participants' estimates of their FLR.