Unsupported score in search output file generated from Peptideshaker and several different search engine output weixiandeng  2019-02-11 10:16
 
Hi Skyline team,

I was trying to build spectrum library through Peptikdeshaker output which is a mzID file, however, it gives me an error report showed below.

Then I switch to comet raw output(pep.xml), tide-search and MSGF+(mzid), they were all given the same error.

Then I tried these files on both Skyline 4.2 and Skyline Daily, still same error.

Can you please help me figure out the problem?

Best,
Weixian

---------------------------
Skyline
---------------------------
ERROR: .mzid file contains an unsupported score type

---------------------------
OK More Info
---------------------------
System.IO.IOException: ERROR: .mzid file contains an unsupported score type

   at pwiz.Common.SystemUtil.ProcessRunner.Run(ProcessStartInfo psi, String stdin, IProgressMonitor progress, IProgressStatus& status, TextWriter writer) in C:\proj\skyline_4_2_x64\pwiz_tools\Shared\Common\SystemUtil\ProcessRunner.cs:line 59
   at pwiz.BiblioSpec.BlibBuild.BuildLibrary(LibraryBuildAction libraryBuildAction, IProgressMonitor progressMonitor, IProgressStatus& status, String[]& ambiguous) in C:\proj\skyline_4_2_x64\pwiz_tools\Shared\BiblioSpec\BlibBuild.cs:line 171
   at pwiz.Skyline.Model.Lib.BiblioSpecLiteBuilder.BuildLibrary(IProgressMonitor progress) in C:\proj\skyline_4_2_x64\pwiz_tools\Skyline\Model\Lib\BiblioSpecLiteBuilder.cs:line 137
---------------------------
 
 
Matt Chambers responded:  2019-02-11 11:59

Hi Weixian,

Such an error on the MS-GF+ mzid files is unexpected: that should work as is. The Comet pepXMLs should be easy to add support for. I'm not surprised about the error on the PeptideShaker and tide-search outputs, and those could require a bit more time to support. Can you upload an example file from each engine for us to test with? Either attach here if it's small enough or https://skyline.ms/project/home/support/file sharing/start.view? - assuming the files are not secret.

Thanks,
-Matt

 
weixiandeng responded:  2019-02-11 13:46

Thank you for your quick response, here are output from MSGF+ and comet. I dropped the file in the pool.

 
Matt Chambers responded:  2019-02-13 08:50

For MSGF+ it seems Skyline needs you to let it do an on-the-fly target/decoy search rather than using a pre-concatenated target/decoy database. That way it can calculate Q-values, and it's Q-values that Skyline is looking for in MSGF+ files.

Just last week they added support for letting you specify an existing decoy prefix: https://github.com/MSGFPlus/msgfplus/issues/58
That will be in release: https://github.com/MSGFPlus/msgfplus/releases/tag/v2019.02.05

Unfortunately that won't work for your database because it doesn't use a prefix, it inserts a string inside the accession.

So there's 3 options for your MSGF+ search:

  1. Redo search with -tda 1 on the target-only database and let MSGF+ generate the decoys.
  2. Regenerate your target/decoy database with decoy prefixes instead of infixes.
  3. Brendan decides BiblioSpec should support multiple scores for MSGF+. However, this would not be as simple as just adding another case to the switch statement which reads scores: it might require treating MSGF+ as 2 different types of analyses depending on whether it has Q-values or not.

For Comet I couldn't test because I don't have the mzML/mzXML file which it needs to in order to extract scan times. Can you upload that as well?

I also discovered the bug with PeptideShaker input. That will be fixed soon.

 
weixiandeng responded:  2019-02-13 09:06

Thanks for your response, I will check out the methods you mentioned for fixing MSGF+ issue, and I dropped the MGF file in the pool which I used for comet search.

 
Matt Chambers responded:  2019-02-13 09:14

Unfortunately only mzML/mzXML are supported by BiblioSpec when importing pepXML, partly because that's the most common scenario, and also because it can be difficult or impossible to automatically map unambiguously from a pepXML result back to the MGF spectrum. So I think that's why your Comet import didn't work. I'll take a look at MGF file and see if it should "just work." Adding MGF as a supported extension for pepXML would be easy, but doing so may lead to errors in other cases due to broken result/spectrum lookup.

 
michael plank responded:  2019-02-18 03:00

Hi Skyline team,

thanks for the discussion above. I`m having similar problems to Weixian, in that I get the error message of using an unsupported score type when trying to build libraries from SearchGUI or SearchGUI+Peptideshaker. I tried a couple of different search engines and the only one I got working was X!Tandem when renaming 't.xml' to '.xtan.xml' (as suggested by someone in another post).

Could you maybe provide an overview of which search engines should work for library-build with SearchGUI/Peptideshaker and what we need to keep in mind.

I previously relied heavily on MaxQuant, but with the current problems with MaxQuant output as mentioned in a different post, this doesn`t seem to be an option. Would you mind sharing the MacCoss lab strategy for library generation in case it uses freeware?

Thanks a lot for your help!

 
phains responded:  2019-03-25 14:55

I’m also having this issue with MS-GF+. I am using SearchGUI to run MS-GF+ and took the resultant mzid file into Skyline to make a library. I did not go through PeptideShaker for this test. I made a decoy database using the TPP.

Decoys have to following format;

DECOY21663
RDANSIDFVAGVAATDVGGTGRIFSAFVMDVNQEVGFRGYPSGPAPLLASESPLLRLELADYIGTSAGSPVAAKAGGLIA

At the end of all of that, I still get the same error message as above. I am using MGFs to search.

Suggestions?

On a related note, I thought you stared fully supporting output from PeptideShaker for import into Skyline when making libraries?

Thanks,

Peter

 
Matt Chambers responded:  2019-03-26 12:23

Please remember to post which variant/version of Skyline you're using when having issues.

  1. The MaxQuant library importer can use external spectra for the last couple of Skyline-dailies, so I'm not sure what issues are that were "mentioned in a different post".
  2. PeptideShaker should also work now in the most recent Skyline-daily.
  3. I don't think MS-GF+ from SearchGUI will work without putting it through PeptideShaker, because it needs mzML/mzXML spectra rather than MGF. MS-GF+ itself should run fine on mzML files, so it's really a SearchGUI limitation.
 
Mike S responded:  2019-04-04 10:55

I would definitely be interested in using just the EValue, rather than the QValue for MSGF+ searches. We sometimes search a very small library where the target-decoy approach doesn't make sense and the QValues wouldn't be that well calibrated.

 
Matt Chambers responded:  2019-06-21 09:22

Added in https://github.com/ProteoWizard/pwiz/pull/576 - but only when QValue is not present. So you may have to modify your files if you let MS-GF+ do a target-decoy search but don't want to use the QValues.