Re: [Psidev-ms-dev] mzML 0.99.0 submitted to document process

[Brian P. <bri...@in...>]

Looks like most commenting happens on this list, so here goes:

 

>From the spec:

 

"The mzData format was a far more flexible format than mzXML. The support of
new technologies could be added to mzData files by adding new controlled
vocabulary terms, while mzXML often required a full schema revision. This is
evidenced by mzData still at version 1.05 while mzXML is currently at
version 3.1. However, mzData did suffer from a problem of inconsistently
used vocabulary terms and there appeared several different dialects of
mzData, encoding the same information in subtly different ways. This was not
usually a problem for human inspection of the file, but caused difficulty
writing and maintaining reader software."

 

This is specious.  The fact that mzData hasn't revved only says to me that
it's badly underspecified, which the paragraph in fact goes on to
illustrate.  The occasional revision of the mzXML schema, to my mind,
indicates a well maintained standard*.  A stable schema and evolving
ontology produce as much or more reader/writer code maintenance work as an
evolving schema-only does.  It's not like mzData readers don't have to be
updated every time something gets added to the ontology.  At least with a
schema there are ways to generate code for these kinds of changes
automatically, and to easily validate the results.  Frankly when it comes to
data formats I think the term "flexible" is synonymous for "trouble" -
convenient for the writers, hell for the readers, and often a dead end for
that reason.

 

I really think mzML will just perpetuate the issues mzData presented.
Better we should figure out a way to generate a proper XML schema based on
the ontology document.  The rest of the world uses proper XML, I really
don't see what makes us special.

 

Well, hey, you asked.

 

- Brian

 

*note that most of the mzXML revisions had to do with things like adding
data compression to peaklists.  It wasn't getting banged around every time
somebody came out with a new mass spec, like the ontology will.

 

  _____  

From: psi...@li...
[mailto:psi...@li...] On Behalf Of Eric
Deutsch
Sent: Tuesday, October 02, 2007 3:32 PM
To: psi...@li...
Cc: Eric Deutsch
Subject: [Psidev-ms-dev] mzML 0.99.0 submitted to document process

 

Hi everyone, I am happy to announce that the mzML 0.99.0 specification
document has been submitted to the PSI document process. This is an
important milestone in the completion of mzML, but it is most certainly not
the end of development and feedback.

 

The specification document and all related materials are publicly available
at:

 

http://psidev.info/index.php?q=node/257

 

There are various kits of instance documents, xsds, the controlled
vocabulary, validators, etc. listed at that site. Please examine and
respond.

 

The actual specification document is posted at:

 

http://psidev.info/index.php?q=node/300

 

You may post comments at that site, or you may send them to this list. We
addressed nearly all issues brought up in the preview period in August. The
one main issue that remains unresolved is the problem of cvParams and how to
handle the inevitable scenario of new terms and older software. This is an
important issue. There is a discussion of it in the specification document.
Your input is sought.

 

We encourage you to begin developing (or adapting) software that implements
the format if you are comfortable knowing that there will be changes before
the 1.0.0 release. I believe that it is primarily by attempting to implement
the format that the community will test the format most rigorously and
reveal issues that still need to be resolved; this is far more effective
than gazing at the specification document.

 

Regards,

Eric

 

 

----------------------------------

Eric Deutsch, Ph.D.
Institute for Systems Biology
1441 North 34th Street
Seattle WA 98103
Tel: 206-732-1397
Fax: 206-732-1260
Email: ede...@sy...
WWW: http://www.systemsbiology.org/Senior_Research_Scientists/Eric_Deutsch