How does multiprecursor searching work in ProteinPilot software 5.0?
In highly complex proteomics data, some percentage of MS/MS spectra is chimeric due to the transmission of more than 1 precursor through the MS isolation step, even with a 1 Da window like the one used on TripleTOF® systems. The ProteinPilot™ Software 5.0 release includes a new multi-precursor search capability that enables the identification of multiple peptides from blended MS/MS spectra. This new feature can only be used when searching *.wiff data, as the multi-precursor analysis relies heavily on the analysis of the raw TOF MS data as part of the process.
First, TOF MS data are reconstructed to determine possible precursors (C12 peaks) in the region of the Q1 isolation window (this scales with the size of the isolation window used). Next, during the database search, these multiple precursors are used to match the potential peptides returned from the search (the default is to use the top 5 most intense precursors). During scoring, the software uses a rigorous approach that prohibits the reuse of evidence; the highest intensity precursors match a set of fragment ions, and these ions are then removed from consideration when the next most intense precursor match is scored. Similar to the protein grouping approach used in the software, this ensures that only unique evidence is used to declare a second peptide in the spectrum.
To validate this feature, we tested it on a large number of TripleTOF system datasets. On aggregate, we found that we got approximately a 5% gain in identifications at the peptide and protein level through multi-precursor searching (Figure 1). To determine if you have any cases in your data, sort on the Spectrum tab by the spectrum number (or in the exports) so that the same spectra are grouped together. Then, look for cases where the Observed m/z column has different m/z values and both have high confidence, which means multiple peaks in the Q1 isolation window have been matched.
Figure 1. With the new multi-precursor search feature, a ~5% gain in identifications at the peptide and protein level was observed.
Note that when iTRAQ® reagent quantification is used, any spectra with multiple identifications will be excluded from quantification through the Shared reporter ion logic.