> On Nov 18, 2014, at 5:20 PM, Chris Moth <cmoth08.gmail.com> wrote:
>
> A recent simulation apparently failed because a heme group tethered to a
> histindine residue (via nmr bond restraint) was wrapped into a different
> periodic cell from the histidine. When the following pmemd simulation
> segment started, the "nmr bond" term spiked to several thousand Kcal,
> and protein structure was badly contorted as energy re-equilibrated.
>
> There is scattered information on this kind of problem in the amber
> reflector, but I'd greatly appreciate a quick update in context of
> Amber14, before I attempt to modify some of the PMEMD source code to
> work around the trouble.
>
> 1) Do you agree that the nmr_lib.F90 disrng() function does not consider
> periodic images in its nmr restraint calculation energy?
Yes.
> 2) Do you agree that wrap_to() in pbc.F90 could be modified so that the
> "center of mass" of the tethering protein could be used to move the heme
> group along with it? Slightly Rephrased... Is there any harm in
> skipping the COM calculation of the heme group, and just wrapping it by
> using the COM of the tethering protein?
Maybe.
> 3) Is there a restraint solution that replaces NMR bond restraints and
> handles differing simulation cell spanning correctly?
No.
> 4) Why isn't this more of a routine problem for others in the Amber
> community?
There are 2 simple workarounds:
1. Set iwrap=0 to prevent wrapping -- then you will never switch images like this. As long as you set ioutfm=1 and ntxo=2, the coordinate files (restart and trajectory) will both be written in binary format, so you will not see numbers overflow fixed-width fields.
2. Use cpptraj to reimage the coordinate file before resuming your simulation.
> 5) I suppose one could build a combination histindine+heme residue and
> use that at the position that is causing trouble. But, that feels
> awkward. Any other advice?
Why not just turn off wrapping?
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 Tue Nov 18 2014 - 15:00:02 PST
