[gmx-developers] XML support in GROMACS

[Thomas Steinke]

David,

thanks for the status update.

We are at a decision point how we should design data repositories which 
can manage results of many GROMACS runs for subsequent data mining / 
post-processing. This includes import/export interfaces and templates of 
typical simulation protocols as well. For an automatic workflow 
processing we are in favour of representing the data (structures, 
simulation parameter, key properties) in XML but *not* the trajectory 
files. Using XML for the FF definitions sounds very promising.

I am aware that the definition of a proper ontology may end in a 
"nightmare" if many expert users are involved. On the other hand, XML 
extensions based on CML may offer sufficient flexibility for the future.
Thus, my question regarding XML support is also to ask whether a GROMACS 
core ontology is already formalized. Even a draft would be helpful.

So far,
Thomas


On 2010-03-08 20:06, David van der Spoel wrote:
> On 2010-03-08 19.43, Thomas Steinke wrote:
>> Dear developers,
>>
>> I am contacting you regarding a future XML (CML) support in GROMACS. The
>> background of my question is as follows: GROMACS is one of the key
>> simulation packages in the German MoSGrid project and we are involved in
>> the design of data repositories using GROMACS data.
>>
>> On the GROMACS Developers Projects page I found that "XML-devel" for
>> Preprocessing is assigned for a 4.x release and that a "XML-spec" is
>> scheduled for the 5.0 release.
>>
>> Can somebody disclose detailed information about the status and possibly
>> release dates for XML support in GROMACS?
>>
>> Thanks in advance,
>> Thomas Steinke
>>
> We have long discussed this, close to eight years if memory serves me
> well, but nothing has been done in core gromacs. Some side projects of
> mine that will become part of gromacs 5.0 use libxml2 and custom-made
> XML documents. I did look into CML at the time, but the overlap with my
> needs (representing quantum chemistry data as well as experimental data)
> was not contained in CML.
>
> There is a decision to make linking to libxml2 obligatory after gromacs
> 4.1, and then a port of some gromacs files to xml might be possible. CML
> might be useful for coordinates etc., but we mainly been considering
> using XML for force fields.
>
> We would welcome contributions of code to increase interoperability of
> gromacs. Maybe you can give some more details about your plans?
>