Re: AMBER: predictive NOE generation

From: Nicolas Lux Fawzi <fawzin.berkeley.edu>
Date: Tue, 27 Nov 2007 10:26:48 -0800

Dear Seth
As an amber user (not developer) I am not aware of any way of doing this
directly in amber (anyone else please correct me...)

If you are interested in predicting NOEs from a single structure (with
hydrogens), you can use the back calculation facility of xplor NIH:
http://nmr.cit.nih.gov/xplor-nih/doc/current/xplor/node478.html
This method calculates a relaxation matrix (the idealized physics behind
the NOE) only from a single structure and an approximate "molecular
tumbling" time. One thing you should take good care to do it specify
the degenerate atoms, since pairs to methyl groups, for example, are
going to have more signal partly because of larger number of protons.

This method doesn't take into account a diverse ensemble of structures,
nor does it incorporate the dynamics of the molecule which both
contribute to the NOE. For more information on one implementation of
these aspects, see papers by van Gunsteren (starting with Peter C, Daura
X, van Gunsteren WF. J Biomol NMR. 2001 Aug;20(4):297-310. Calculation
of NMR-relaxation parameters for flexible molecules from molecular
dynamics simulations. ---- and subsequent papers). This work is a
(more) general case of the calculation performed in xplor nih and
reading that paper might help you understand what is being calculated in
xplor nih.

I have some forthcoming work for a peptide system which closely follows
this above implementation (except the simulations I did were run in
sander and pmemd and most inputs to the predictions were calculated with
ptraj) and enables very good prediction of Overhauser effect
"enhancements" (ie cross peaks). I would warn though that for any
system with significant flexibility or conformational diversity, this
full treatment will only be useful if you tightly converge the sampling
the equilibrium ensemble as well as make a good estimate of the
individual hydrogen time correlation functions (dynamics).

If all of the above is too complicated if all you want is a simple
"check", many people will say to check if the r^-6 averaged pair
distance is within 5 or 6 angstroms, which is most of the time a
necessary (but not sufficient) condition for a true NOE. I would be
very careful with that and perhaps account also for degenerate protons
here as well by calculating intensities instead of just averaged distances.

Please let me know if you think I could be of more assistance!

-Nick



Seth Lilavivat wrote:
> Dear Amber Users,
>
> Is there a way to generate hypothetical NOE's from a model? I would
> like a pointer to a good reference if someone has had success with
> this.
>
> Thanks,
> Seth
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber.scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
>
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Wed Nov 28 2007 - 06:07:38 PST
Custom Search