When to use Combination searches
Most database searches are either Peptide Mass Fingerprints, with a list of peptide molecular mass values, or MS/MS searches, where the mass of each peptide precursor is only used to filter candidate sequences from the database before scoring matches for the fragments. It is also possible to combine the two, in a so-called Combination search, where both the molecular mass data and the fragment mass data are scored in a single result report. Often, this is chosen for data from a TOF/TOF instrument, where molecular mass information can be acquired more rapidly than MS/MS.
You can only view the results of a Combination search in a Protein Summary report. If you switch to any of the other report formats, the PMF scoring is discarded and the results are a conventional representation of the MS/MS matches.
One situation where a Combination search can be useful is if you have high coverage PMF data for a protein, plus a limited amount of MS/MS data. The PMF score contribution may be similar to the sum of the scores from the MS/MS spectra, and combining the two can give increased confidence in the protein identity. In particular, if the MS/MS matches are from peptides shared between many entries in the database, the PMF information can help decide which of these proteins is the preferred hit.
Click on this link for an example. The single MS/MS spectrum gives a match to LQIWDTAGQER with a score of 64. This peptide, with four permutations of L and I, is found in 25 human Ras-related proteins in SwissProt, and there is nothing in the result report to distinguish between these possibilities. In fact, if you switch to the protein family summary, you’ll find all 25 matches collapse into a single family. Switch to the Protein Summary (ignoring the warning that it is deprecated) and you’ll see that the molecular mass information strongly favours RAB1A_HUMAN, where the PMF is worth an additional score of 65. For RAB1B_HUMAN, a very homologous protein, the PMF score drops to an additional 46 and for the others it is a negligible 22 or less.
One word of caution. There is danger in including very poor quality MS/MS data in a Combination search. Imagine that an MS/MS spectrum has a precursor mass match to a protein, but the MS/MS spectrum is mainly noise and doesn’t give a match to the peptide sequence. We must then say that this mass does not belong to the protein, so there should be no contribution from the peptide mass to the PMF score. In other words, bad MS/MS data can degrade a good PMF.
The main limitation of a Combination search is that of any PMF: it will only work for a single protein or a very simple mixture, maybe 2 or 3 proteins of similar abundance. If the mixture is more complex than this, or if the protein of interest is a minor component, the contribution from the PMF score will be negligible. This is not a problem as long as you also look at the results in one of the MS/MS summary report formats (Peptide, Select, or Protein Family). Trying to view MS/MS data from a mixture of any complexity in the Protein Summary is misleading, because it is limited to 50 hits, and minor components may be pushed off the list by variants of major components.
Typically, a spectrum used for a peptide mass fingerprint has similar information content to a single MS/MS spectrum. Maybe 50 monisotopic peaks at a given mass accuracy. So, if your instrument gives similar amounts of each, and you are analysing single proteins or simple mixtures, such as individual spots off a 2D gel, Combination searches can be very useful. But, always remember to look at the results in an MS/MS format also, to see the protein inference based on the MS/MS matches, and make sure you aren’t missing a minor component.
Keywords: combination search, PMF