It's hard to say for certain without reading your scripts, but are  
you sure
that the square well width (in Hz) that you're using with the  
potential is
less than the tolerance (in Hz) above which you declare a particular  
Jcoup
restraint to be violated?

Also note that even if your violation tolerance is greater than
your squarewell width, you can still introduce errors.  That's  
because the
forces created by the Jcoup potential depend on the magnitude of the
difference between the calculated coupling and the edge of the tolerated
region, ie., Jobserved +/- squarewell-width.  So if your squarewell  
width
is 0.5 Hz, and a particular restraint is violated by 0.6 Hz, the  
atomic forces
are only 3% of what they would be if you'd used the harmonic potential.
So even if though it's violated, that Jcoup restraint is not pushed  
back into
place very hard.

Generally speaking, you should keep the Jcoup potential as harmonic.
Problems in your structure are far more likely to be caused by mis- 
assigned
NOEs than by 3J couplings.

--JK

On Aug 14, 2007, at 12:23 PM, Andrew Borgert wrote:

> I am running an SA script in vacuum on a small glycopeptide, and I
> noticed something strange (to me at least) happen when I changed the
> j-coupling potential type.  I initially ran my script using a harmonic
> potential for the backbone j-coupling ( C-N-CA-C ), however when I
> switched to a square potential, the number of j-coupling restraint
> violations over the ensemble of structures increased dramatically  
> (by a
> factor of about 2) while the number of violations of other types of
> restraints (noe, dihedral, bond, angle etc.) stayed the same.  This
> seems somewhat counter-intuitive, as switching to a square potential
> should be akin to 'loosening' the restraints a bit, which should  
> result
> in fewer violations.  Any thoughts?
> I realize that my description of the problem is rather vague, so  
> I'm not
> looking for specific trouble shooting, just general advice.
> I am running xplor_nih 2.12.
>
> Thanks
> Andrew Borgert
> Biophysical Sciences and Medical Physics
> University of Minnesota
>
> _______________________________________________
> Xplor-nih mailing list
> Xplor-nih at nmr.cit.nih.gov
> http://dcb.cit.nih.gov/mailman/listinfo/xplor-nih

Reply via email to