OpenMS/OpenMS

The codebase of the OpenMS project

Timo Sachsenberg

@timosachsenberg

in the mzIdentML...

this is wrong in several ways: 1. the ontology (cvRef) is unimod not PSI-MS

and I think it should be 2. name="Carbamidomethyl"

but I will check how this usually looks like

        <SearchModification fixedMod="true" massDelta="57.021464000000002" residues="C">
            <cvParam accession="UNIMOD:4" cvRef="PSI-MS" name="Carbamidomethyl"/>
        </SearchModification>

so maybe I was wrong and PSI-MS is ok as cvRef

but the name needs to be Carbamidomethyl

Probably should be fixed in metamorpheus

Timo Sachsenberg

@timosachsenberg

can work around that issue by doing a search and replace and see if it works?

Miguel Cosenza-Contreras

@MiguelCos

I tried modifying the mod files in metamorpheus by defining the modification as "Carbamidomethyl" instead of Carbamidomethyl on C. But the problem persists. I will give another try and let you know.

jpfeuffer

@jpfeuffer

@timosachsenberg If PSI-MS is okay as cvRef, then our code is wrong: https://github.com/OpenMS/OpenMS/blob/03fda657546db090a1dca5812609261dd49e988f/src/openms/source/FORMAT/HANDLERS/MzIdentMLHandler.cpp#L348

You have to do a double search-replace

Miguel Cosenza-Contreras

@MiguelCos

Just ran a second try, same error. Apparently MetaMorpheus is not actually allowing me to change the names of the modifications or the way it writes them. I might as well try to just "create a new modification" and try.

jpfeuffer

@jpfeuffer

Are you able to modify the mzID file with Unix tools? Then you could do a simple sed

Timo Sachsenberg

@timosachsenberg

@jpfeuffer I think the cvRef here is ignored

but it takes the accession and extracts the cvref from there

at least that would explain why our test files also have PSI-MS

and we don't get any error there

jpfeuffer

@jpfeuffer

Well SearchModification is ignored completely anyhow

you have to look for Modification tags

probably on the sequences themselves? I didnt look into the file yet

https://github.com/OpenMS/OpenMS/blob/develop/src/tests/class_tests/openms/data/MzIdentML_XLMS_labelled.mzid#L71

I assume XL has special code anyway

or is it using the DOMHandler?

Timo Sachsenberg

@timosachsenberg

good point I will check

<Peptide id="P_7">
  <PeptideSequence>ALANVNIGSLICNVGAGGPAPAAGAAPAGGPAPSTAAAPAEEK</PeptideSequence>
  <Modification location="12" residues="C" monoisotopicMassDelta="57.021463721">
    <cvParam name="Carbamidomethyl on C" cvRef="PSI-MS" accession="UNIMOD:4" />
  </Modification>
</Peptide>
<Peptide id="P_8">

Miguel Cosenza-Contreras

@MiguelCos

Are you able to modify the mzID file with Unix tools? Then you could do a simple sed

I don't know how to do that but I can definitively can try.

4 replies

Miguel Cosenza-Contreras

@MiguelCos

Are you able to modify the mzID file with Unix tools? Then you could do a simple sed

This work-around worked. I opened the mzID on Notepad++ and manually deleted all the " on M"; " on U", etc... Then the pepXML file was produced and it is not empty.

Many thanks for the input!

M Bernt

@bernt-matthias

I have a little (Galaxy workflow) that runs MSGFPlusAdapter -> PeptideIndexer -> FalseDiscoveryRate ; which fails with

FalseDiscoveryRate failed due to missing information:
It seems like protein inference was not yet performed. Calculating Protein FDR is probably not meaningful. To override, use the force flag.

Does anyone have an idea how this can happen?

2 replies

Timo Sachsenberg

@timosachsenberg

I think we need to disable protein:FDR calculation in that case in the FDR tool

emilpaulitz

@emilpaulitz

Hello everyone,
I have a problem building OpenMS on my Windows 10. I tried to cautiously follow the instructions on https://abibuilder.informatik.uni-tuebingen.de/archive/openms/Documentation/release/latest/html/install_win.html, but when giving the final build command after installing the contribs and qt, it gives the following error:

CMake Error at C:/Program Files/CMake/share/cmake-3.18/Modules/FindPackageHandleStandardArgs.cmake:165 (message):
Could NOT find XercesC (missing: XercesC_LIBRARIES XercesC_INCLUDE_DIRS)
Call Stack (most recent call first):
C:/Program Files/CMake/share/cmake-3.18/Modules/FindPackageHandleStandardArgs.cmake:458 (_FPHSA_FAILURE_MESSAGE)
cmake/modules/FindXercesC.cmake:82 (find_package_handle_standard_args)
src/openms/cmake_findExternalLibs.cmake:56 (find_package)
src/openms/CMakeLists.txt:70 (include)

The error log refers to another error, which appears in the console log as follows:

-- Performing Test COMPILER_HAS_DEPRECATED_ATTR
-- Performing Test COMPILER_HAS_DEPRECATED_ATTR - Failed
-- Performing Test COMPILER_HAS_DEPRECATED
-- Performing Test COMPILER_HAS_DEPRECATED - Success

Which is why I rather suspect the Xerces-C installation to be the problem. There are around 1000 lines of error messages in the xerces-c section of the contib_build.log similar to:

xerces-c_3.lib(ICUTransService.obj) : error LNK2001: unresolved external symbol u_strlen_58 [C:\path\contrib_win64_build\src\Xerces-C_3_2_0\samples\Redirect.vcxproj]

I tried re-installing the contribs, and installing xerces-c manually from the website and integrating the resulting folders to the contribs folder structure, with unchanged results. Any ideas for this particular error?

Timo Sachsenberg

@timosachsenberg

can you post how you configured / built the contrib ?

did you open a 64bit command prompt

emilpaulitz

@emilpaulitz

I opened a VS 2019 64 bit prompt with the vcvars64.bat, as in the tutorial, and then used
"cmake -DBUILD_TYPE=ALL -DNUMBER_OF_JOBS=4 -G "Visual Studio 16 2019" -A x64 "C:\path\OpenMS\contrib"

jpfeuffer

@jpfeuffer

can you upload the full contrib-build.log, specifically the CMake configuration of Xerces

you are probably missing a unicode transcoder

emilpaulitz

@emilpaulitz

contrib_build.log

I re-built only xerces a couple of times, so this is the only content of the log. Thanks for the replies btw!

jpfeuffer

@jpfeuffer

it finds the ICU from your conda environment

is a conda env activated?

emilpaulitz

@emilpaulitz

No, and I tried it again after specifically deactivating conda, unchanged result.

jpfeuffer

@jpfeuffer

you have to delete the whole xerces folder in the contrib build first

it needs to reconfigure itself

src/xerces-3.2 or so

emilpaulitz

@emilpaulitz

Yes, I deleted everything with Xerces-C something. It still failed building Xerces, so I temporarily removed the unicode folders from my two conda environments, so cmake could not find them. It then said "Failed to find all ICU components", but the installation went through without errors, until the OpenMS tests all just gave Exit code 0xc0000142.
So I downloaded the latest ICU4C and added the directory to my path, so it was found by cmake and the installation worked until building OpenMS in VisualStudio. This gave the unresolved external symbol errors with Xerces-C again. Are the failed tests due to another issue or should I do something else regarding the ICU problem?

jpfeuffer

@jpfeuffer

you do not need ICU. Xerces uses the system transcode if it is not found.

jpfeuffer

@jpfeuffer

I think the failing tests are just pointing to another problem.

jpfeuffer

@jpfeuffer

Could be Virus check (https://community.adobe.com/t5/adobe-media-encoder/0xc0000142-error-on-adobe-media-encoder/td-p/10066152?page=1&profile.language=de) or User permissions or corrupted file system. For virus check, disable the Update check.

Also rule out missing system dependencies after building by running DependencyWalker on the test executables: https://github.com/lucasg/Dependencies

Where communities thrive

People

Repo info

Activity