| Brendan MacLean responded: |
2018-12-16 17:49 |
Hi Srikanth,
Can you provide the full log? It feels like there must be more to this error. If there is no other error message from BlibBuild or BlibFilter, then it would seem to be a failure simply running one of these EXEs.
What version are you using? We did just make some dependency changes to BlibBuild to support reading spectra from the mass spec raw data files for MaxQuant instead of using the embedded spectra in the msms.txt files, but those would only be in Skyline-daily.
If you are using Skyline-daily, then it would be useful to know whether you can achieve your desired pipeline with the current Skyline 4.2 release.
Thanks for posting to the Skyline support board.
--Brendan
|
| |
| smanda responded: |
2018-12-17 17:09 |
Hi Brendan,
I have tried BlibBuild separately on the same file, and get the error:" ERROR: .mzid file contains an unsupported score type". The Skyline version is 4.2.0 18305.
The PeptideShaker version is 1.16.29.
Where can I get the full log?
Regards,
Srikanth
|
| |
| Brendan MacLean responded: |
2018-12-17 18:37 |
Hi Srikanth,
Super strange. Skyline should be able to parse that message from BlibBuild and display it in the message box by virtue of it starting with "ERROR:"
So, I am kind of curious why you don't just see exactly the same message when building in Skyline. If you are not using "English (US)" language settings on your computer, could you switch and see if you get the expected error message instead of the more generic error about running the process failing?
I am just looking for things that might be unique about your system that would cause it not to behave as I would expect. Certainly, many other people have reported this type of message without first seeing the process failed message.
Let's keep trying to make progress on that issue, but also please post the file you are trying to build from to:
http://skyline.ms/files.url
And we will have a look at why the score type is not being detected. I am pretty sure we have seen PeptideShaker mzIdentML before. But, regardless, I am sure we can have this working in a Skyline-daily soon. A Skyline major release may take longer.
Thanks for using BlibBuild to understand better what is going on. Hopefully, we can fix both the score support and the less than helpful error message.
--Brendan
|
| |
| smanda responded: |
2018-12-18 15:49 |
Hi Brendan,
I am using English (US), I tried on windows 7 and also Windows Server and get the same error. I have uploaded the mzid output of combined search engine export from PeptideShaker as Combined.7z to the portal.
I re-exported the results just to be sure, things not wrong on PeptideShaker output export.
Regards,
Srikanth
|
| |
| smanda responded: |
2018-12-18 15:51 |
Hi Brendan,
In order to skip the peptideshaker step, I see that I can build a spectral library directly from the outputs of the search engines using the Blibuild. But not sure, if there if it applies any kind of FDRs on the merging?
Regards,
Srikanth
|
| |
| Nick Shulman responded: |
2018-12-19 02:02 |
Thanks for sending us that file.
The reason that you are getting that unsupported score type error is that the first SpectrumIdentificationResult's in the .mzid file have a "Mascot:expectation value" score. Because of this, BiblioSpec expects all of the SpectrumIdentificationResult's to have that type of score.
However, the eighth SpectrumIdentificationResult in the file (on line 3456436) does not have a Mascot expectation value. It has a couple of other scores that would have been valid, but at that point BiblioSpec is only looking for a Mascot expectation value.
It sounds like we might need to relax the assumption in BiblioSpec that all of the spectrum id's in the file will have the same score types.
-- Nick |
| |
| smanda responded: |
2018-12-19 17:12 |
Hi Nick,
Thanks for the quick response. Can I expect a new Daily with the changes?
It worked for me with the earlier versions, does it mean you have made changes to newer version of BiblioSpec?
Regards,
Srikanth |
| |
| Brendan MacLean responded: |
2018-12-19 18:19 |
Hi Srikanth,
It likely means you did something different with this file. I very much doubt this exact file worked with earlier versions. Though, you can always install older versions from the "Unplugged" installers page:
https://skyline.ms/wiki/home/software/Skyline/page.view?name=install-64-disconnected_4_2
Just click "I Agree" and then "Archive" and choose the version you believe could handle this file. Let us know if you find one.
Otherwise, we need to know how this file came into existence. It seems like some combination of previous results files of various types? Some from Mascot and some not?
Please explain and explain how you would expect us to resolve the fact that the same score types are not present on each spectrum match. This is something we do not feel we have seen before and something BiblioSpec is not yet designed to handle.
No Skyline-daily fix is imminent, because it would likely take some rethinking of BiblioSpec to support this, and we aren't yet sure what the use case is and whether we can simply come up with a workaround that would allow you to build a library.
Thanks for helping us understand your use case.
--Brendan |
| |
| smanda responded: |
2018-12-19 19:09 |
Hi Brendan,
When I mean it worked with earlier versions, I did not mean with same file. The approach was similar but with a different dataset altogether.
My workflow involves:
Search mgf files using Mascot, X!Tandem and MSGF+ and then combine the resulting files using PeptideShaker at 1 % peptide/protein FDR. Then I export the results to mzid format. I was importing this into Skyline to create the final spectral library. I use Skyline to calibrate using RT peptides and then export the results in openswath format report.
Let me know your thoughts over this workflow or if you can suggest an alternative.
I am aware that I can directly import all the results into BiblioSpec and build the library, but afraid it does not do anything for the FDR control.
Regards,
Srikanth |
| |
| Brendan MacLean responded: |
2018-12-20 11:16 |
Hi Srikanth,
If you still have a file that you used this workflow on to build a library successfully, please try that same file again with the current build of Skyline, and I would expect that to work again, just as I would expect your current file to fail with older versions. I don't think we have changed much in this area.
If you could provide an mzIdentML file that worked, that might be instructive and help us figure out what the important difference is between that file and the one you have provided.
Thanks for helping us understand better.
--Brendan |
| |