Exploring science is typically characterized by a lot of puzzles, frustrations or even failures. This weblog is mainly intended to record my working, thinking and knowledge acquisitions. I expect that some reflection would refresh my mind from time to time, and motivate me to move further, and hopefully give me a better view about even changing the landscape of bioinformatics.
You are welcome to leave some comments, good or bad, but hopefully something constructive. Enjoy your surfing!
Monday, December 6, 2010
mass accuracy of Orbitrap: internal communications
From communications with one colleague.
It was mentioned during the training course last month that, while performing DDA, the Orbitrap performs a "pre-scan" of the ions entering the mass spectrometer. This pre-scan is really just a part of the full scan. Rather than waiting for the completion of the full scan at a high resolution, instead, part way through, it registers the pre-scan (at a resolution of about 15,000) for the purpose of selecting ions for MS/MS. The full scan continues to completion, but the MS/MS scans on the linear ion trap already are underway by the time this happens. The point here is that the precursor masses reported for the MS/MS spectra are from the pre-scan, not the higher resolution (if one was specified in the methods file) full scan. At my request, he installed a utility named extractMSn on the computer associated with OT1. This utility takes a .RAW data file and extracts all of the MS and MS/MS spectra, creating a set of .dta files. The first line of each .dta file representing a MS/MS file contains the m/z value of the precursor as determined by the full scan, not the pre-scan.
The purpose of the pre-scan is to speed up the process of data-dependent acquisition (DDA). Instead of waiting for an entire full scan to be completed before doing the first MS/MS, it collects data when the full scan is partially complete. This doesn't mean it only has covered a part of the scan range. The full range of ions are present in both the pre-scan and the full scan; they are just better resolved in the full scan (and a little more accurate).
The pre-scan is not inherently more useful than the full scan. In fact, the peaks of the full scan should be better resolved. The problem is that the precursor m/z values reported for the MS/MS spectra (i.e. the values that appear in the .dtas) are the less-resolved pre-scan values. So something is needed to go back to the full scan and pick out the better resolved values if those are what one wants.
I decided to have a look at whether it makes any signficant difference. So, for 63 MS/MS spectra representing BSA peptides, I calculated the ppm error for precursor m/z values taken from the pre-scan and the full scan. This data is attached as a spreadsheet. For the pre-scan precursor m/z values, the ppm error ranged from -2.1 ppm to as high as 10.9 ppm in what seems to be a uniform distribution. For the full scan precursor values, most fell into the range -2.8 ppm to 3.4 ppm, with an average of -0.8. This matches what I've observed for y ions of MS/MS spectra. However, there was a second range or errors, from -31.9 ppm to -22.1 ppm, for the precursors of 11 peptides. Possibly, the second range arises from the utility selecting an interfering peak rather than the proper one in certain instances. I'll look into it further.
That is what ExtractMSn does (generates peak lists with the full scan m/z value for the precursor included instead of the pre-scan value), but it apparently doesn't always do it well. If it did it with 100% success, I would recommend using it for applications where resolution greater than 15,000 is important. But it doesn't.
In any case, it seems the extractMSn utility isn't very helpful and should not be used.