Spectral library building from MaxQuant msms.txts

support
Spectral library building from MaxQuant msms.txts danielz  2018-07-13 05:56
 

Dear Skyline team,

foremost thank you very much for providing such a tool to the community.
I have a suggestion on the spectral library building from MaxQuant output files (that I hope was not raised too often before):

MaxQuant does a lot of peak processing on the original raw scan, i.e. it tries to decharge multiply charged fragment ions if also a 1+ version of the fragment is detected. This does basically no harm to 2+ precursors, but for precursors with z>= 3+ a lot of times it means that the spectrum that is stored in the MaxQuant .res files as well as all fragment m/z information that is written as annotated peaks to the msms.txt file does not represent what the raw scan looked like.
In an extreme example, after building a library from the msms.txt and there is a theoretical 3+ peptide, there is now an y12(1+) fragment at 1201 m/z ion in the spectral library which is the base peak for that spectrum. If Skyline now trys to extract that transition from the PRM/SRM/MRM raw file you will have tough luck, because that fragment ion is actually present at 600 mz as y12(2+). Ergo: Low dotp, lower confidence, lower signal.

As the decharging of spectra cannot be disabled in MQ, my current workarounds are a) use Mascot/Sequest/EngineXYZ where you can control this parameter (by then potentially imparing the DDA data search through noisier and more complex MS2 spectra). Or b) you extract the raw scans yourself, manually or using some implementation of the ThermoRAWFile Reader library.
Spectronaut handles this by talking the spectrum number and rawfile name from the IDs in the msms.txt and goes back to the raw file (if path provided) to extract the raw scan, annotates the scan and stores that it in the spectral library.

My question is a) are you aware of this, b) is this recognized as a problem and c) are there plans to integrate such functionality or is the direct import of spectral libraries from MaxQuant files anyway only used by a low proportion of users, because most spectral libs come from different tools/sources?

Best,
Daniel

 
 
Brendan MacLean responded:  2018-07-13 06:13

Hi Daniel,
Yes, we are aware of the problem, but your request helps to keep this an important issue for us. Please make the same request from the MaxQuant team. Although, I have already forwarded this to Juergen, who is also aware of the issue and said the MaxQuant team would fix it several times over a number of years.

It seems it may finally be time for us to give up that hope and go with the back-up solution of requiring the original spectrum files to be present for MaxQuant results, as is necessary with other formats, like pepXML. It is unfortunate, because this will slow library building down and generate new support requests as people struggle with this extra requirement, but it may be a necessary stop-gap until we can get the MaxQuant team to add the necessary raw spectra into the msms.txt.

Thanks for bringing this up again. Sorry, we have let it slide for so long.

--Brendan

 
danielz responded:  2018-07-13 06:27

Hi Brendan,

thanks for the prompt response. Good to know that this is being worked on. In order be independent of code changes to MaxQuant one could take a look at http://planetorbitrap.com/rawfilereader which at least under Windows allows for the fast extraction of the raw files directly. I have no clue about licensing with Thermo etc, but the raw file reader libraries work pretty performant in our in-house implementation (used for another case where we need raw spectra).
I will drop a line at the MQ people as well, perhaps it helps.

Best,
Daniel

 
Brendan MacLean responded:  2018-07-13 06:49

Hi Daniel,
Don't worry. We are very good at reading mass spec formats and all 6 major instrument vendors (including Thermo) fund Skyline and take an active interest in our being able to read their latest file formats. There is no technical barrier, just one of prioritizing work and determining the best solution, which would still be to have MaxQuant put the raw spectra in the msms.txt either instead of the deconvoluted spectra or with them, but if we can't get the MaxQuant team to do that, then we just need to do what we are doing for other search formats in finding the spectrum source files and reading directly from them. Probably not that much work.

Again, thanks for bringing it up.

--Brendan