Unsupported score in search output file generated from Peptideshaker and several different search engine output

Unsupported score in search output file generated from Peptideshaker and several different search engine output weixiandeng  2019-02-11
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?


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.chambers42 responded:  2019-02-11

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.


weixiandeng responded:  2019-02-11

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

matt.chambers42 responded:  2019-02-13

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

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.chambers42 responded:  2019-02-13

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

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

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;


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


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



matt.chambers42 responded:  2019-03-26

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

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.chambers42 responded:  2019-06-21

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.