Now that I think about it, 'autoimage' should work for bilayers
reasonably well. I suggest trying the "autoimage" command as Dan
suggested first (but it may be worth understanding the scheme I outlined
below since it might help understand periodic boundary conditions a bit
better).
Good luck,
Jason
On Mon, 2014-03-10 at 11:00 -0400, Jason Swails wrote:
> On Mon, 2014-03-10 at 14:18 +0000, Vijay Manickam Achari wrote:
> > I have simulated bilayer in water. There are 128 lipid in total and
> > around 600 water molecules on top and bottom of the lipid bilayer.
> >
> > Up to about 50 to 60 ns the trajectory works fine. Then I find the
> > water molecules shift to one layer more than the other layer.
> > After that I also notice the layers in the bilayer shift upside down.
> > The water molecules become in the center and the tails of each bilayer
> > pointing upward and bottom of the bilayer.
>
> This is just an imaging artifact, as it seems that you suspected.
>
> >
> > I used mdinput as below:
> >
> >
> > %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
> > Production 300ns constant pressure & zero surface tension
> > &cntrl
> > imin=0,
> > irest=1, ntx=5,
> > ntt=2, temp0=355.0,
> > ntxo=1, iwrap=1, nscm=5000,
> > ntb=2, pres0=1.0,
> > ntp=3, csurften=3, gamma_ten=0.0, ninterface=2,
> > ntc=2, ntf=2,
> > nstlim=100000, dt=0.002,
> > ntpr=100, ntwr=-50000, ntwx=1000, ntwe=500,
> > cut=10.0, ioutfm=1
> > /
> > %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
> >
> >
> > Since I used "iwrap" command, I used "unwrap" command and reimaged the
> > trajectory using commands as below:
>
> The "unwrap" command does nothing here since the 'reimaging' simply
> reverses the unwrap action. [1] Since the 'unwrap' command seems to be
> computationally demanding, I suggest getting rid of it.
>
> > %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
> > trajin ../$system-MD300-run0900.traj 1 100 5
> > trajin ../$system-MD300-run1000.traj 1 100 5
> >
> >
> > unwrap :1-1214
> > center :1-1214
> > image center
>
> You want "image origin" here. The "center" command sets the origin to
> the center of geometry (or mass) of the mask. The "center" keyword in
> the image command uses the center of mass of each molecule when imaging
> instead of the first atom. I typically use "image origin center"
> myself.
>
> > trajout abc.traj mdcrd
>
> I'd suggest using NetCDF as a format -- it's better in almost every way
> (higher precision, much smaller file, and harder to corrupt).
>
> In general, if you make sure that your analyses use the minimum image
> convention when computing distances (as cpptraj does), it doesn't really
> matter if your trajectory is not imaged the way you like it.
>
> If you want the bilayer to 'look nice', then you need to define a box
> center that will roughly correspond to the middle of the bilayer. The
> image command will then grab all of the periodic images that are closest
> to that point. If your bilayer extends to the sides of the box (as I
> suspect it does), then you can simply center the system on an atom that
> resides at the very end of a _single_ lipid tail. Try this:
>
> ======
> trajin your_trajectory
>
> center :<RESIDUE>.<ATOM>
> image origin center
>
> trajout abc.nc netcdf
> ======
>
> You'll need to replace <RESIDUE> with the number of a residue that
> represents the tail of one of your bilayer lipids. <ATOM> is the atom
> name of the atom at the end of the lipid chain.
>
> I think (and hope) this will work, but I've never actually run any
> bilayer simulations.
>
> Good luck,
> Jason
>
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Mar 10 2014 - 08:30:02 PDT
