Forums: Define.xml
Hello Sherry,
Discussing with our P21 define team, please see the follow-up to each question below:
Kind regards,
Matt
As one of the developers of the Define-XML standards (versions 1.0, 2.0 and 2.1), such postings make me pretty sad ...
Personal opinion: If software claims to generate compliant define.xml files, it should:
- not be based on "naming conventions" for OIDs (Object Identifiers): the user should be able to use whatever they want (e.g. UUIDs)
- allow to add comments (references to def:CommentDef) everywhere this is allowed, without exception
- not change or "adapt" any information like adding underscores or so
- not remove any information provided by the user like for the "arm:AnalysisDatasets" element, or refuse to allow to provide it. Every element of the Define-XML standard should be fully supported.
- support all elements and attributes (like "Purpose") for all versions of all CDISC submission standards
- also support generation of metadata for tabular datasets that are not based on submission standards ("generic" use of define.xml)
- not make any assuptions regarding what the user wants to accomplish, like that the define.xml is in the context of a regulatory submission. The users should be able to decide these things themselves.
There are very good software packages on the market that are fully define.xml compliant, e.g. that use graphical user interfaces for adding all the necessary and allowed information (i.e. not based on Excel "specifications") in a very user-friendly way, help the user making decisions through "wizards", have pre-defined templates for the most common submission standards and versions (but with the possibility to change everything, including OIDs), have build-in validation against XML-schema and other rules, generate the HTML view of the define.xml "on the fly", and much much more.
These software packages are not always free of charge, but have none of the limitations mentioned in Sherry's posting, and have excellent customer support.
Once again: all this is my personal opinion.
Thank you for pushing v2.1 functionality also into Community version.
I had tried as suggested in your presentation to take CDISC's sample define.xml in the v2.1 release package and to create the P21 spec off it to see an example what needs to be filled in.
Out of curiousity, I've loaded the P21 spec back into P21 to generated the define.xml again to see if it matches with CDISC's release but there are some differences:
1. For the AdaM define.xml, in the MetaDataVersion, although multiple def:Standard were available, but the def:CommentOID="COM.STDCT.01" is not included (although def:CommentDef is present).
2. In the arm:ResultDisplay/Name attribute, it adds in underscores. E.g.,
"<arm:ResultDisplay OID="RD.Table_14-3.01" Name="Table 14-3.01">"
becomes
"<arm:ResultDisplay OID="RD.Table_14-3.01" Name="Table_14-3.01">"
3. The def:PDFPageRef/Title attribute seems to be completely dropped off (but it is actually one of the features in v2.1), therefore it used the def:title element instead to populated. E.g.
"<def:PDFPageRef PageRefs="2" Type="PhysicalRef" Title="Table 14-3.01"/>"
becomes
"<def:PDFPageRef Type="PhysicalRef" PageRefs="2"/>"
4. The elements in <arm:AnalysisDatasets> seems to be completely taken out.
5. For the SDTM define.xml, the Purpose attribute under ItemGroupDef seems to be missing if using SDTMIG-MD / AP.
Lastly, I also have the folloiwng question:
1. Where can the def:Context attribute be defined? I can seem to find where I can populate this as 'Other'. It is always 'Submission'.
Thank you.