Re: [AMBER] Reg: Query on distance restrain and its effect on conformation

From: Jason Swails <jason.swails.gmail.com>
Date: Wed, 11 Mar 2015 11:41:12 -0400

On Wed, 2015-03-11 at 00:41 -0400, GG Smith wrote:
> Dear Amber Users,
>
> Am a graduate student venturing recently in to dynamics, I
> would like to humbly seek your advice and suggestion regarding a
> simulation attempted by me on Protein+nucleic acid complex .
>
> I had written to amber forum about the ideal range force constant to
> be used for restraining distance b/w specific residues of protein &
> nucleic acid backbone. I was suggested to use ~ 20 kcal/mol. Am using
> AMBER12 (ff12SB) program.
>
> Pls. find below the steps i followed:
>
> *Am trying to position a nucleic acid duplex with enzyme by imposing
> (Guiding) a specific distance between them (between side chain of
> certain amino acid and phosphate backbone atom.
>
> *The starting distance between them was ~ 7 Angstrom and i wanted it
> in range of ~5.0 A.
>
> This is my restraint file;
>
> # 16 ASP OD2 42 DNA O1P
> &rst
> ixpk= 0, nxpk= 0, iat= 292, 1317, r1= 4.30, r2= 4.80, r3= 4.90, r4= 5.40
> rk2=20.0, rk3=20.0, ir6=1, ialtd=0,
> &end
>
> *During equilibration process of free energy minimization, i imposed
> this condition. I was able to achieve it to near value by setting rk2
> & rk3 to 20 kcal/mol. I continued it in pre-equil production run and
> then for 5ns of production run. But during last 1 ns i see that some
> Watson & crick paired bases are not maintaining planarity & at times
> losing hydrogen bond and duplex structure distorted a bit. I have a
> doubt that whether this is due to continued restraint during
> production run MD too.
>
> I have done MD simulation for the same system without restraint and
> the duplex looks fine and the hydrogen bonds between WC paired
> bases are maintained.

This suggests to me that the problem could be with the restraint. 20
kcal/mol is a fairly strong restraint (although weaker than most bonds).
At a distance of 2 Angstroms from equilibrium distance, this introduces
a 80 kcal/mol/A force on each of the two atoms (in different
directions). To see how this compares to "typical" forces, attached is
a histogram of the magnitudes of the forces on every atom in a
particular (well-equilibrated) snapshot from one of my own simulations.
As you can see, there are *some* forces larger than 80 kcal/mol/A, but
very few. In fact, in this 20K atom system (using explicit solvent with
PME), only 38 atoms had forces this large (about 0.2% of all atoms in
the system). And I suspect this is a sustained force (i.e., it remains
a large force for several steps).

Think about what this restraint is doing -- you are attaching a spring
between two individual atoms so that the entire force of the restraint
is applied to those two atoms. Try and picture how pulling a floppy
tube-like structure through water by pulling hard on a single point
could distort the structure of that molecule.

It is fairly easy to rationalize why the force could be the cause of the
structure distorting (it doesn't mean that this is definitely the
culprit, but it seems to me to be a sensible hypothesis). So I would
suggest looking for a way to be gentler about moving the NA duplex where
you want it. Some ideas:

- Try pulling in steps. You can use Steered MD to move the target
distance during the course of the simulation (see jar=1 in the reference
manual). Or you can run this stage in multiple steps, each one setting
the distance closer to where you want it.
- Try using center of mass restraints or multiple distance restraints in
order to distribute the force more evenly over the duplex.

This might take some playing around to get things positioned the way you
want them. Of course, there's always the possibility that this is
simply what happens when the NA and protein come this close in those
starting configurations. In that case, you'll need a different approach
altogether.

HTH,
Jason

-- 
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher



_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber

unknown-2L9EVX
(image/png attachment: unknown-2L9EVX)

Received on Wed Mar 11 2015 - 09:00:05 PDT
Custom Search