Re: [Psidev-ms-dev] Current mapping file

[Lennart M. <len...@eb...>]

Hi Marc,


> thanks for the up-to-date link.
> The other version I found is at: 
> http://psidev.cvs.sourceforge.net/psidev/psi/psi-ms/mzML/schema/ms-mapping.xml?view=log 

Well spotted. I have now deleted this from CVS, and remarked in the log 
comment of the commit that the files should be obtained from the psidev 
SVN instead, along with the path to the validator configuration files.


> I have several questions/comments regarding the mapping file.
> 
> Comment 1:
> There is no mapping for the "/mzML/run" tag

Correct, I couldn't easily find anything that we require or expect 
there, so I let the rule out. Maybe there should be one or more rules 
added for this location, but fleshing out the mapping file is still a 
work in progress. You're very welcome to make suggestions, actually.


> Comment 2:
> There is no mapping for the 
> "/mzML/run/spectrumList/spectrum/spectrumDescription/precursorList/precursor/isolationWindow" 
> tag

Correct, we should add a meaningful rule here.


> Comment 3:
> There is no mapping for the "data processing parameter" terms (MS:1000630)

Correct again; needs to be added.


> Question 1:
> According to the mapping file the "ms level" term (MS:1000511) should be 
> under "spectrumDescription", but is under "spectrum" in the tiny1 example.
> Which one is right?

Good question. Can't remember off the top of my head what this should 
be. So you can see: there's work yet to be done on the mapping file!


Essentially, my main effort in these matters right now is spent on 
getting our reference validator up and running. Once that is achieved, 
the mapping file can be refined as we go along (the whole thing is 
descriptively configured from the XML file, so no code needs changing).

I think we should probably have a conference call dedicated to the 
mapping file at a point in the near future, but that is open to discussion.

Regardless, it would be good to have the reference validator out there 
first, so people are free to play with the mapping rules and instance 
documents. At the EBI, we're ironing out the last bugs in the validator 
now (while the mzML validator itself is simple and consists of nearly 
trivial code, this is only because (i) we have this rather complex but 
minimal memory footprint jmzML component that effectively uses an mzML 
file as a cache file, and (ii) we build the entire back-end on the 
generic PSI validator framework, which is a very complicated suite of 
modules that are high-tech codepieces in their own right -- this 
framework is also used by the PSI-MI people, for instance).


Cheers,

lnnrt.