On Dec 3, 2007 10:47 PM, Gary Thompson <garyt@xxxxxxxxxxxxxxx> wrote:
on an aside should we start supporting the ccpn datamodel?
Support of the CCPN data model (http://www.ccpn.ac.uk/ccpn/data-model)
can of course be added to relax, and would be a welcome addition.
Well for the 1.3 line anyway. Can you envisage a practical use for
this? Is there a specific application? Do you mean for the structure
reader? Or for the spectral side of things?
both... It should be possible to set things up so you extract relaxation
parameters from the datamodel and return the results back to the
datamodel...
Is the datamodel really ready for that? Last we talked to Tim & Rasmus
about this, there was nowhere in the datamodel for results of a
model-free analysis. That was a long time ago, but I get the
impression they have been busy with things other than that - correct
me if I'm wrong. It raises a few curly, if pedantic, issues: eg. to
what element of the molecular system does an order parameter belong?
In the case of amide relaxation, it should be the amide bond vector,
in the case of methyls, it belongs to the symetry axis. Are these even
meaningful concepts in the datamodels representation of the molecular
system? (It becomes even more tricky if we consider relaxation of
nuclei with non-colinear sources of relaxation!)
On a more parctical note, I wouldn't be encouraging anyone to use
Analysis to determine relaxation rates. Neither the peak integration
nor rate-fitting routines were up to the job, last I played with them
(Again, this may have changed recently, but I've missed it in the
release notes if it has).
None of my negativity is meant to imply that this is a bad idea,
however. It would be great if Analysis and the data model were up to
suplying relax with data, and vice-versa. I'd just point out it there
is more work to be done than just some import/export functions in
relax (otherwise I might have done it myself 18 months ago!)
Cheers,
Chris
Re: [Fwd: Re: CCPN data model.]