# Re: applying experimental constraints to a MD simulation

From: Jerome Henin (jhenin_at_cmm.chem.upenn.edu)
Date: Fri Mar 16 2007 - 11:57:25 CDT

Hi Michel,

If you apply a harmonic restraint on the distance with force constant k at
temperature T, you should get an approximate Gaussian distribution of width
sigma = sqrt( RT / k )
(it will be rigorously Gaussian if the restraint force is much stronger than
whatever coupling there is to other dgrees of freedom).

So there is no need to define an interval: you should be able to tune your
force constant directly based on the experimental dispersion.

Best,
Jerome

On Friday 16 March 2007 12:31, L. Michel Espinoza-Fonseca wrote:
> Hi all,
>
> Below find my original posting. Dan suggested a very interesting
> approach to my problem (restraining the distance between two atoms) by
> using the free energy module of NAMD. The approach sounds interesting
> and I think it can be potentially used to solve my problem. However, I
> have one more question. I have a very specific distribution of
> distances obtained from experiments (let's say, x +/- 2*stdev). Such
> distribution accounts for 95 % of the population, based on our
> Gaussian distributions. The problem here is that (If I'm right) if I
> apply this:
>
> dist ( PEP1,6,ca) (PEP2,6,ca) kf=20 ref=13.337
>
> Then I won't have a real control on my distance distribution, is that
> correct? So I looked at the manual, and I found out that I can do the
> following:
>
> dist ( PEP1,6,ca) (PEP2,6,ca) kf=20 low=13.337 hi=8.337
>
> So the distances fluctuate within these two values. I'm not very
> familiar with this procedure, so I'd like to know if I'm correct. If
> not, I'm open for suggestions!!!
>
> Thanks!
> Michel
>
> ORIGINAL POSTING:
>
> Hi all,
>
> I was wondering if my specific problem can be solved with NAMD -I
> really hope so. Right now, we have all this bunch of experimental data
> (CA-CA distance distributions, to be more exact), and I'd like to use
> them as "constraints" to build a 3-D model of a protein, using MD
> simulations. I've used the harmonic constraints parameters included in
> NAMD before, but usually you apply a force instead of a given
> distribution of distances. Based on this, I'd like to know if any of
> you have an idea on how to fit one (or multiple) distribution(s) of
> distances (let's say, from 45 to 55 A) into the MD simulation. All
> comments will be greatly appreciated!!!
>
> Thanks!
> Michel
>
> 2007/3/12, Dan Bolintineanu <bolintineanu_at_cems.umn.edu>:
> > I think you can use the urestraint command, which is part of the free
> > energy calculation module. Something like:
> >
> > freeEnergy on
> > freeEnergyConfig {
> > urestraint {
> > dist ( PEP1,6,ca) (PEP2,6,ca) kf=20 ref=13.337
> > dist ( PEP1,6,ca) (PEP2,8,ca) kf=20 ref=13.3244
> > dist ( PEP1,6,ca) (PEP2,13,ca) kf=20 ref=9.65392
> > }
> > pmf {
> > task = stop
> > lambda = 1
> > time = 260 ps
> > }
> > }
> >
> > Not to worry, this won't actually carry out free energy/pmf-type
> > calculations, it's just the syntax required to make it do what you
> > want. Change time to however long your simulation is (you'll need to
> > specify numsteps > time somewhere else in the config file). kf is an
> > effective force constant (the higher this is, the more rigid your
> > constraint), and ref is the target restraint distance. The selection
> > I have there is for atom-to-atom distance restraints (segid, resid,
> > atom name), but you can do much more sophisticated ones
> > (centre-of-mass to centre-of-mass, angles, etc.). See "Options for
> > confromational restraints" in the manual (in my version, it's section
> > 6.7.3).
> > Good luck.

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:44:28 CST