Re: [AMBER] Calculating electrostatic potential in cpptraj

From: Jason Swails <jason.swails.gmail.com>
Date: Sat, 16 Aug 2014 21:05:24 -0400

On Sat, Aug 16, 2014 at 4:20 PM, Covington, Cody Lance <
cody.l.covington.vanderbilt.edu> wrote:

> Thanks Jason
>
> >> Hello all
> >>
> >> Is there a way to calculate the electrostatic potential at a given
> >> atom position or point in space in cpptraj?
> >
> >
> >cpptraj cannot (yet) do PME electrostatics. There are a couple actions
> that do electrostatic
> >calculations, but they do direct space pair-pair interactions only using
> minimum image convention.
> >
>
> What are the methods to calculate through the direct space sum?
>

​The "pairwise" and "lie" actions compute electrostatic and van der Waals
interactions (as does the 'spam' action). I think only "pairwise" will
really give you what you want.


>
> >
> >>
> >> Or any other analysis program?
> >>
> >
> >None in Amber that I know about.
> >
> >
> >>
> >> I can calculate it using my own programs, but I want to use PME to get
> >> long range electrostatics that would go beyond where periodic
> >> boundary conditions end. This might seem silly, but I have a
> >> surfactant system that has a high ionic strength.
> >>
> >>
> >
> >
> >>
> >>
> >> I have also tried to use
> >>
> >> image xoffset 0 yoffset 0 zoffset 0
> >>
> >> image xoffset 0 yoffset 0 zoffset 1
> >>
> >> ... etc
> >>
> >> To make the system more like and 'infinite' one, but this is quite
> >> cumbersome.
> >>
> >
> >It will be _very_ hard to make the system large enough that you can
> compute an electrostatic potential
> >using just a direct sum-over-pairs (and it will get _very_ slow very
> quickly). Your best bet really is
> >to use an Ewald-based method. There are a couple options:
>
> I have done some tests and I think I can use the direct space sum if I can
> get about 20 nm between my structure and the boundary. The potentials are
> different but the fields are about the same.
>

Well the field strength drops off as 1/r^2, whereas the ESP varies as 1/r,
so that would make sense. But you had said you wanted the electrostatic
potential, not the electric field. And the field is easier, anyway -- just
compute the electrostatic force at a particular point and divide by the
charge (no need to take any kinds of differences to isolate the ESP at a
single point from an energy).

Thanks for the suggestions, I also wondered if I could somehow use amber's
> PME code somehow?
>

I suppose so -- calculating the ESP may actually be easier with Amber now
that I think about it so long as you can add the dummy particle (but
calculating the electric field I think would be harder). You'll need to
create an additional topology file with the extra dummy particle (assuming
you want the ESP and not just the electric field). ParmEd _can_ be used to
add the extra particle, but it would probably be easier to use tleap.

Then compute the energy using sander or pmemd using a single-point energy
calculation. Extract the electrostatic energy (EEL and the 1-4 EEL values
added together) from the output file and take the difference in the way I
described before.

The electric field would be harder to get, since you would need to get
_just_ the electrostatic force (which is not decomposed like the energy is
in the output file). You would need to use ParmEd to zero-out all of the
bond, angle, torsion, and Lennard-Jones terms, then compute and print the
force for each particle using the "ntwf" variable.

HTH,
Jason

-- 
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Aug 16 2014 - 18:30:02 PDT
Custom Search