[pride-support #1213] Issue 5 in pride-converter-2: Can't figure out what the mzIdentML parser requires...

[GoogleCodeExporter]

What steps will reproduce the problem?
1. Attempt to convert an mzid file generated by a user-script

Example problematic mzid file available on request.

What is the expected output? What do you see instead?

Lack of compliance to the PRIDE converter 2 assumed instantiation of mzIdentML 
is not surprising, but the error message provided (Failed to unmarshall 
dbSequence: ... at MzIdentmlDAO.java:1496) masks the true exception from 
convertIdentification. I expect to see the true exception, rather than the 
catch all exception I do see.

What version of the product are you using? On what operating system?

2.0.12, Windows. 

Please provide any additional information below.

I'm generating mzIdentML from third-party PSM data-formats using custom scripts 
and a C++ program that links with the ProteoWizard libraries. I've been able to 
work around various surprises (spectrum identifiers) in the supported version 
of mzIdentML, but can't get past the exception above - which provides no 
information about what needs to be tweaked.

Please unmask the exceptions from 
MzIdentmlDAO.java::MzIdentmlDAO::convertIdentification! 

Original issue reported on code.google.com by edward...@gmail.com on 10 Dec 2012 at 6:44

[GoogleCodeExporter]

Hello Edward,
thanks for reporting this bug. We have changed the source code to attach the 
original exception to the one thrown by the DAO. The next converter version 
will come with this change. We usually don't release a new version for minor 
bugfixes like that. You could build the converter yourself or we could build a 
SNAPSHOT version for you.
Best Regards,
the PRIDE team

Original comment by florian....@gmail.com on 11 Dec 2012 at 9:09

• Changed title: [pride-support #1213] Issue 5 in pride-converter-2: Can't figure out what the mzIdentML parser requires...
• Changed state: Started

[GoogleCodeExporter]

Not sure where the SNAPSHOT is, so I'll try to do a build. I'll have to install 
maven too - seems straightforward enough. Is there a build document somewhere?

Original comment by edward...@gmail.com on 11 Dec 2012 at 8:03

[GoogleCodeExporter]

OK, got a lot of errors when I tried "maven compile" in the directory 
pride-converter. I tried the same thing in dao-mzidentml and it seemed to 
build. Not sure how to build the entire thing. Advice?

Original comment by edward...@gmail.com on 11 Dec 2012 at 9:32

[GoogleCodeExporter]

Building the converter may not be straight forward if you are not familiar with 
Maven. I would have to see the errors you get to tell you more about what needs 
to be done. (One problem you will have is that some dependencies may not be 
provided automatically due to licensing)

As an alternative, you can just replace the mzidentml DAO. Once you have 
unpacked the converter archive, it will have produced a 'lib' directroy. 
Replace the provided dao-mzidentml jar file in that folder with your newly 
build one (keeping the same file name as the provided one) and the converter 
should use the new code.

Original comment by florian....@gmail.com on 12 Dec 2012 at 10:57

[GoogleCodeExporter]

This is what I did. I used mvn package in the dao-mzidentml and tweaked my 
mzIdentML converter to fix (most) of the issues I identified or commented out 
the problematic element of the p-c-2 code. I still have other issues to solve, 
but this one at least, can be put to bed.

The issues I identified with mzIdentML parsing are:

Assumption of a AnalysisProtocolCollection element - proteowizard doesn't 
output this element if there are no SpectrumIdentificationProtocol elements 
inside. 

Assumption of start and end positions for peptides in the PeptideEvidence 
element.

Original comment by edward...@gmail.com on 13 Dec 2012 at 5:15