Error building spectral library from TPP/PeptideProphet pep.xml

support
Error building spectral library from TPP/PeptideProphet pep.xml david schmidt  2019-11-06
 
Hi,
I'm trying to import search results produced by X!Tandem and further analysed by PeptideProphet using the TPP, in order to build a spectral library in Skyline. The original data was acquired using a DDA approach on a waters QTOF, centroided and lockmass-corrected in MassLynx and then converted to MzML using MSConvert. Importing the MzML as results works fine. However, when trying to build a spectral library from either the interact.pep.xml or the tandem.pep.xml file, I'm getting the error: No spectra were found in the new library (see attached).

I suspect that this may be due to the link within the search results that point to the spectra in the MzML file. A pretty similar problem was described before, but I couldn't figure out what the solution to that was:
https://skyline.ms/announcements/home/support/thread.view?rowId=3056

Thank you very much for your help,

David
 
 
matt.chambers42 responded:  2019-11-06
Hi David,

What version of Skyline[-daily]?
 
david schmidt responded:  2019-11-06
Hi Matt,

19.1.0.193
 
matt.chambers42 responded:  2019-11-06
Was the X! Tandem search done on the mzML you created with msconvert?
 
david schmidt responded:  2019-11-06
Yes. I don't know whether the one can move the mzml and pep.xml files into different paths before importing to skyline.
 
matt.chambers42 responded:  2019-11-06
No, you had better leave them where they were when you did the search. Or have the mzMLs and pepXMLs in the same directory.
 
david schmidt responded:  2019-11-06
I did exactly that as the first try. I tried both, leaving them where they were (in different directories) as well as moving them into the same. In addition, moving both into the skyline working directory also didn't help. But if you think that the path is indeed the problem, I could re-run the TPP pipeline and write results into the same path.
 
matt.chambers42 responded:  2019-11-06
Hmm, can you upload the mzMLs so I can try to reproduce?
 
david schmidt responded:  2019-11-06
I'll try but I'm afraid that's too big (even zipped it has 190 Mb).
 
david schmidt responded:  2019-11-06
Yes, the upload indeed failed. I'll re-run the TPP analysis tomorrow and will try again.
 
matt.chambers42 responded:  2019-11-06
If the data's NOT confidential you can upload it to https://skyline.ms/project/home/support/file%20sharing/begin.view? - if it is confidential and the TPP re-run still fails, we'll find another way to transfer.
 
david schmidt responded:  2019-11-06
It is not confidential. I uploaded the zipped version (191008_Hela_Lysate_STD_2AFAMM.zip).
 
david schmidt responded:  2019-11-07
Hi,

I just found out, that the import of search results from the TPP Comet pipeline (based on the exact same MzML file and further processed with PeptideProphet) into Skyline works. Are there certain changes to be made for X!Tandem based input?
 
matt.chambers42 responded:  2019-12-03
Hi David, just got to look at this. The interact.pep.xml for Waters mzML input is plain wrong. It uses the scan number from the Waters id as the overall scan number, but each Waters function has its own set of scan numbers. So there can be "function=2 scan=23" and "function=3 scan=23" and they're totally different scans. But in the interact.pep.xml there is no way to distinguish them, it just says "23". Comet's pepXML output works because they write the full nativeID ("function=1 process=0 scan=234"). If you need to use this particular search pipeline you'll need to go through a "lowest common denominator" format, mzXML probably best. That will reduce the nativeIDs to monotonically increasing "scan numbers" at the cost of them no longer being directly connected to the raw Waters spectrum ids. But you should be able to feed that mzXML to any TPP tool and Skyline should read it the same way and be able to find ids.
 
david schmidt responded:  2019-12-04
Hi Matt,

thank you for the clearification. I actually just found out that both TTP pipelines and the import into Skyline work when I perform peak picking and centroiding/lock mass correction in PLGS 3.0.3, export as MGF (somehow the MzML files exported directly from PLGS are missing an index offset required by the search engines) and then convert to MzML using MSconvert. Quite a complicated workflow but at least it works. Thanks a lot for your support!

Cheers,

David