On Fri, Apr 17, 2015 at 5:23 AM, Hannes Loeffler <Hannes.Loeffler.stfc.ac.uk
> wrote:
> Hi,
>
> I'm interested in getting the ligand-solvent interaction energy from a
> periodic, explicit solvent simulation. I know I can use the
> subtraction method using sander -y and imin=5 and recruiting cpptraj
> with strip/parmstrip to do the same for the parts (solvent and ligand).
>
> I would like to get this energy through the direct method, i.e. by
> computing the non-bonded energies between ligand and solvent directly.
> It looks like this is possible through sanders LIE facility and this
> also seems to be supported with imin=5.
>
This can't really be done with PME, since part of the electrostatic isn't
done directly. The best you can hope to do is the 3-step method you
described using sander's imin=5 and cpptraj parmstrip/strip.
cpptraj's "lie" action does a direct calculation, but it uses an energy
shifting function to solve energy discontinuities at the cutoff. sander's
LRT functionality does 3 PME calculations each time LRT energies are
requested -- one with the solvent charges 0, one with the solute charges 0,
and the full electrostatics; the EEL energy is full-solvent0-solute0. The
vdW energy is computed directly. Of course this has the benefit of *not*
requiring 3 trajectories and prmtops and 3 sander simulations, but it's
still doing the same amount of calculation.
I get a final result in the last two lines of the output
>
> Final LIE Evaluation:
> Evals: 10 E(LRT): -3.26 RMS 1.29 E(Disp): -9.96 RMS
> 1.00 SASA: 247.52 RMS 1.58
>
> I assume E(LRT) is EE (actually 0.5 of the interaction EEL) and E(Disp)
> is VWD?
Yes.
> I am only interested in the average energy. Judging from an
> old email in the archive
>
> http://archive.ambermd.org/201402/0019.html
>
> this result is for full PME and long-range corrections so I would
> expect this to be directly comparable to the subtraction method.
>
I have the same expectation. Though I've never verified myself. A quick
glance at the code is enough to convince me that the EEL energy is
correct. I can't say the same about the vdW, though (but just because I
haven't dug through the code deep enough). You can always test a couple
frames with cpptraj's LIE, since the vdW part should be done equivalently
in both sander and cpptraj (i.e., it is computed directly).
All the best,
Jason
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Apr 17 2015 - 06:00:03 PDT
