AMBER Archive (2007)

Subject: Re: AMBER: torsion angle penalty calculationy

From: Mike Summers (summers_at_hhmi.umbc.edu)
Date: Wed Jan 03 2007 - 12:13:24 CST

Thanks, David. This was very helpful.

Mike

On Wed, Jan 03, 2007 at 08:54:33AM -0800, David A. Case wrote:
> On Wed, Jan 03, 2007, Mike Summers wrote:
> >
> > I'm trying to fix torsion angles using torsion restraints
> > in an NMR restraint file.
> >
> > FROM THE RESTRAINT FILE:
> > #####################################
> > # C6-C7-N1-C8
> > &rst
> > rk2=10000.0, rk3=10000.0, ialtd=0,
> > iat=2273,2274,2263,2275, r1=-180.10, r2=-180.05, r3=-179.95, r4=-179.90, /
> > #####################################
> >
> >
> > When I look at the output file, it appears that the penalty
> > for the torsion angle restraints is being calculated differently
> > from the NOE-distance penalties, even though IPNLTY is set to 1.
>
> First, the manual is misleading about IPNLTY: this applies only to NOESY
> volume or chemical shift restraints. It is ignored for distance and torsion
> restraints: there the potentials listed on pp. 180-181 are applied. I will
> prepare an erratum to explain this correctly.
>
> >
> > Restraints, deviations, and energy contributions: pencut = 0.00
> >
> > ------------------------------------------------------------------------------
> > First atom Last atom curr. value target deviation penalty
> > ------------------------------------------------------------------------------
> > C7 cap 146 -- N1 cap 146: -180.156 -180.050 0.106 0.025 t
> >
> > I assume it is the penalty that is used during the structure calculations.
> > If so, is there a flag or parameter that allows the penalty for torsion angles
> > to be calculated linearly with deviation, or better yet, to scale at a much
> > greater rate than linear? The torsion angles are varying significantly during
> > the calculations, and I'd like to keep them relatively rigid.
>
> It looks to me like the program is doing the right thing: in the example
> above, you have a deviation of .106 degrees = .0018 radians. The parabolic
> penalty is then 10000 * (.0018)^2 = 0.034 kcal/mol. Since the current
> value is actually below r1, the potential flattens out (to linear, rather than
> parabolic), so it is even smaller that the parabolic value would have been.
>
> I think what you want to do is to make the difference between r1 and r2 bigger
> (also between r3 and r4). Since you want to keep the torsions rigid, try
> setting r1 to -200 (say) and r4 to -160. This will make the penalty bigger
> for larger deviations. For your case, a 1 degree deviation would then have
> a 3 kcal/mol penalty, and a 5 deg. deviation would have a 76 kcal penalty, so
> you should not see deviations larger than a few degrees.
>
> [It is also no problem to set r2=r3; then you have a simple parabolic
> potential = rk2*( curr - 180)^2, which is easy to calculate by hand to check
> what the program is doing.]
>
> If you need to constrain things even more tightly, you would have to increase
> the values of rk2 and rk3 accordingly. But you might run into problems with
> the MD algorithm if the constraint gets really strong.
>
> ...good luck....dac
>
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu 

-- 

*********************************
Michael F. Summers
Department of Chemistry and Biochemistry
  and Howard Hughes Medical Institute
University of Maryland Baltimore County
1000 Hilltop Circle 
Baltimore, MD 21250

Phone: (410)-455-2527  
FAX:   (410)-455-1174
Email: summers_at_hhmi.umbc.edu
Web:   www.hhmi.umbc.edu
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu