Hi Jakob,
Are you using the current version of xplor-nih? I recall Charles
made some recent corrections to his scripts because of errors with
the ASSIgn...OR... syntax.
The ASSIgn...OR... syntax for ambiguous NOE restraints is poorly
documented. Michael Nilges (I believe) added it to the NOE term back
in the xplor 3.8 days, where it supported an early form of ARIA.
It's described in the output of "NOE HELP", but no where else. The
way it works is fairly obvious: Only the atom pairs selected by the
paired selections in each clause are included in the distance
averaging for the restraint. The idea is to prevent irrelevant atom
pairs from being included when you're expressing an ambiguous
restraint. The original NOE ASSIgn syntax would often end up
including atom pairs that you didn't intend to have considered. For
example, if a particular peak had three possible assignments:
from A to B
from C to D
from E to F
and you expressed it as
ASSIgn (A or C or E) (B or D or F) 3.0 1.2 1.0
the distance averaging would extend over the atom pairs
from A to B
from A to D
from A to F
from C to B
from C to D
from C to F
from E to B
from E to D
from E to F
which is obviously wrong. Expressing this ambiguous restraint as
ASSIgn (A) (B) 3.0 1.2 1.0
OR (C) (D)
OR (E) (F)
ensures that the distance averaging only extends over the three
correct atom pairs.
Note that if you have extensive ambiguity in your NOE dataset, you're
probably better off using my Marvin scripts to calculate structures,
since they're far more error-tolerant than the classic xplor NOE term.
Let me know if you still have questions or problems.
--JK
On Dec 10, 2007, at 7:51 AM, Jakob Toudahl Nielsen wrote:
> Dear all,
>
> I am testing a refinement protocol using ambiguous distance restraints
> editing the "~/eginput/gb1_rdc/refine.py" xplor-nih python script
> template.
> Most of my ambigous distance restraints contain the OR syntax (here
> ambiguous hydrogen bonds):
> -------------------------------
> assign (resid 14 and name O) (resid 2 and name HN) 1.00 1.00
> 1.00 OR
> (resid 15 and name HN) (resid 2 and name O) OR
> (resid 14 and name O) (resid 3 and name HN) OR
> (resid 15 and name HN) (resid 3 and name O) OR
> ...
> ...
> -------------------------------
> The files pdbTemplatename.viols and averageFilename.viols reports much
> violations that expected. I use the xray-structure as starting
> structure which should have only 10 violation but end up with 48
> violations after refinement. Inspection of the .viol files reveals
> that the violations are exclusively reported with respect to the first
> sele pair in the ambioguous distance restraint. I.e. violation for the
> distance between 14O and 2HN in the example above regardless if 15HN
> and 3O fullfill the distance restraint.
> I end up with high noe violation energy (3689 with scale 30 and soft
> with asympSlope=0.2)
> I do not specify the setAvetype("sum") - this is the default - right?
> I only observe very few violations for the unambiguous distance
> restraints
>
> It seems like either:
> 1) I am missing something obvious - sorry if this is the case!
> 2) The analyze procedure does not recognize the OR syntax.
> 3) The OR syntax is not recognized at any stage.
>
> What do you think is the problem?
>
> Any help wpuld be very much appreciated!
>
>
> Best regards,
>
> Jakob
>
> ----------------------------------------------------------------
> This message was sent using IMP, the Internet Messaging Program.
>
> _______________________________________________
> Xplor-nih mailing list
> Xplor-nih at nmr.cit.nih.gov
> http://dcb.cit.nih.gov/mailman/listinfo/xplor-nih
