Dear Eldon Ulrich,
You may remember me from the ENC conference earlier this year in
Monterey, California. We talked about making it easier for users to
submit relaxation data and model-free analysis results to the BMRB,
possibly by taking the XML results files from the program relax
(http://nmr-relax.com), or by relax producing STAR formatted output
specifically designed for the BMRB (the better option, as database
fields can be made mandatory). I hope that you are still interested
in this BMRB-relax integration. And I also hope that you don't mind
that I have added the relax development mailing list to the cc field.
relax is developed as community project and is fully open source
(https://gna.org/projects/relax/, one of the only programs in NMR
being so), and hence the other relax developers are likely to be
interested in these developments.
Since that time, I have been busy updating the program relax to be
more suitable for the task of interfacing with the BMRB deposition
system. That is the reason it has taken me so long to contact you.
There is a development branch of my program in which I have rewritten
almost the entirety of the program, for reasons other than BMRB
integration, which is where I plan on adding this support. This
branch has been a long time in development but should soon be usable
for model-free analysis. Recently I have added the feature that
model-free results (or results from any type of dynamics analysis for
that matter) can be saved to an XML formatted file. An example of
this is the highly truncated OMP model-free results in the file
'final_results_trunc_1.3.bz2 ' which is viewable in the directory
http://svn.gna.org/viewcvs/relax/1.3/test_suite/shared_data/model_free/OMP/.
As this is a development branch of relax, this file format is likely
to change from time to time.
The easiest path for relax integration with the BMRB would be to first
add all the mandatory BMRB fields into relax, outputting the relavent
results into a special XML formatted file, and then finally writing
some simple STAR writing code (unfortunately Python has built in XML
support but no STAR format creation or parsing capabilities) to
produce the format of
http://www.bmrb.wisc.edu/dictionary/htmldocs/nmr_star/dictionary_files/nmrstar3.txt.
I'm not sure if this format is yet fully capable of storing all of
the information from a complete model-free analysis, and it is highly
tailored to Art Palmer's Modelfree program. For example in relax you
can optimise the full double motion model of Clore et al., 1990, and
hence you simultaneously have both tf and ts parameters (Dasha has the
same capabilities). There also appears to be no way of storing the
essential diffusion tensor information. The field
_Order_param.Model_free_sum_squared_errs is also not quite right as
this is really a chi-squared value rather than SSE value (it's the SSE
divided by the individual errors to give the SSE normalised to unit
variance - the chi-squared value). And additional fields allowing the
users of BMRB to judge the quality of the collected relaxation data
would also be very powerful to differentiate useful data from the
chaff. If you plan to be at the ICMRBS conference in August, I will
also be there if you would like to talk more about this integration.
Regards,
Edward
P.S. The addition of MPI or grid computing capabilities to run on
clusters, etc. will soon be added to this branch of relax as well.
This will enable server integration and open up the possibility of
users submitting relaxation data to a web server (consisting of a
cluster or supercomputer) and having the final results returned to
them in relax or BMRB format, if this is of any interest.
Relaxation and dynamics: interfacing the program relax with the BMRB.