A uniform proteomics MS/MS analysis platform utilizing open XML file formats

Abstract

The analysis of tandem mass (MS/MS) data to identify and quantify proteins is hampered by the heterogeneity of file formats at the raw spectral data, peptide identification, and protein identification levels. Different mass spectrometers output their raw spectral data in a variety of proprietary formats, and alternative methods that assign peptides to MS/MS spectra and infer protein identifications from those peptide assignments each write their results in different formats. Here we describe an MS/MS analysis platform, the Trans-Proteomic Pipeline, which makes use of open XML file formats for storage of data at the raw spectral data, peptide, and protein levels. This platform enables uniform analysis and exchange of MS/MS data generated from a variety of different instruments, and assigned peptides using a variety of different database search programs. We demonstrate this by applying the pipeline to data sets generated by ThermoFinnigan LCQ, ABI 4700 MALDI-TOF/TOF, and Waters Q-TOF instruments, and searched in turn using SEQUEST, Mascot, and COMET.

Figure 1

Trans-Protoemic Pipeline using open XML file formats at three steps: (1) raw spectral data generated by different mass spectrometers; (2) peptide assignments using different search engines; and (3) protein identifications using different methods of inference. Asterisk indicates that PeptideProphet must be specialized for each search engine.

Figure 2

Trans-Protoemic Pipeline analysis of LC-MS/MS data sets. (A) Accuracy of PeptideProphet-computed peptide probabilities for HaloICAT LCQ data set in sliding window of 50 search results. (B) Numbers of search results for HaloICAT LCQ data set filtered at a minimum PeptideProphet probability to achieve a predicted 2.5% error rate. The inset shows the numbers using Mascot results with probabilities adjusted by SearchCombiner to take into account the results of SEQUEST and COMET applied to the same data set. (C) Numbers of ProteinProphet identifications for HaloICAT LCQ data set filtered at a minimum ProteinProphet probability to achieve a predicted 2.5% error rate. Each asterisk indicates an incorrect protein identification. (D) Numbers of ProteinProphet identifications for Serum MALDI-TOF/TOF data set filtered at a minimum ProteinProphet probability to achieve a predicted 2.5% error rate. (E) Numbers of ProteinProphet identifications for Yeast Q-TOF data set filtered at a minimum ProteinProphet probability to achieve a predicted 2.5% error rate.