On Wed, Oct 21, 2015 at 5:13 AM, Karolina Markowska <markowska.kar.gmail.com
> wrote:
> 2015-10-19 15:16 GMT+02:00 David A Case <dacase.scarletmail.rutgers.edu>:
>
> > On Mon, Oct 19, 2015, Karolina Markowska wrote:
> > >
> > > I have a problem with my system. It's a small globular protein (286
> aa).
> > > When I was doing short simulations (10 ns) there was no problem, but
> > right
> > > now, when I want to do 50 ns simulations (5 x 10 ns) I see there is
> > > something wrong with the system.
> > > The volume is getting smaller, density is increasing and the total
> energy
> > > is decreasing very fast.
> >
> > How much (numerically) is the density increasing? Are you starting the
> > 50ns
> > simulations from the end of the 10ns run? Are you using sander or pmemd?
> > What do you see when you visualize the trajectory? Look *really*
> > carefully to
> > see if there are any differences between the 10ns simulations and the
> > 5x10ns
> > simulations.
> >
> > Now I see that there is also something funny with the density. It keep
> increasing from 1.03 up to 1.053 and then it decrease.
> I'm starting a new run. I'm not using the old 10ns simulation files. I'm
> using pmemd.CUDA.
>
Why is this surprising? It's not unusual for system densities to be this
high for protein systems (or even higher).
Well, when I use this script:
>
> reference protein_complex.inpcrd
> trajin prod1.traj.gz
> trajin prod2.traj.gz
> trajin prod3.traj.gz
> trajin prod4.traj.gz
> trajin prod5.traj.gz
> autoimage
> center :1-286 mass origin
> image origin center familiar
>
These last two commands, "center" and "image" are probably redundant.
"autoimage" usually does what you want it to.
> rms ToRef :1-286.CA,C,N= reference out rmsd.arg mass
> atomicfluct out rmsf.arg :1-286 byres
> trajout protein_f.binpos binpos
> go
>
> everything looks fine.
> But when I get rid of autoimage, center and image commands, I see that my
> water box is spinning over the center of the protein and the protein is
> sticking out of the water box.
>
This is an imaging artifact, as proven by the fact that "autoimage"
removes this problem. It has no effect on the simulation itself.
> I think that this spinning octahedral could be something funny.
>
I really don't know what this means... it doesn't make sense to me that
autoimage could remove box rotations (the unit cell should not rotate --
the unit cell vectors should be fixed in direction through the whole
simulation, and only the lengths allowed to change if using a barostat).
Unless I'm missing something...
> But probably this is not the main problem. The problem is, I think, that I
> probably have free spaces between water from 3D-RISM and tleap waterbox.
> I can see them, whet I make a pdb file from my .prmtop and .crd files from
> each step of simulation. After minimization I've got this free spaces, but
> during heating and equilibration everything looks OK to me.
The barostat will press out these empty spaces, and it should be fine. In
fact, tleap puts a lot of space between the edge of the box and the closest
atoms, so there is going to be fairly large gaps if you visualize the
system replicated in all directions. This leads to a *very* low internal
pressure and simulation boiling if you run at constant volume. But using a
Barostat will decrease the volume of the system until you force out these
vacuum bubbles.
> And when I try
> to make a .pdb file from production coordinates file, I see that the
> waterbox looks quite normal, except it have a hole on one side, and my
> protein sticking out form another side of the box.
> That's kinda weird.
>
Again, it's an imaging artifact.
> I think, I should change something during minimization.
> I've found this script from D. SIndhikara's tutorial:
>
> % minimization of solvent orientation
>
> &cntrl
> ntmin = 1,
> imin = 1,
> maxcyc = 1000,
> ncyc = 250,
> ntb = 1,
> igb = 0,
> ntr=1,
> restraint_wt=10.0,
>
> restraintmask='(:WAT.O) & ((!:WAT) <:5)',
>
> ibelly=1,
> bellymask=':WAT',
>
This looks weird to me. Typically you use *either* restraints *or* belly,
but not both (and I would suggest using restraints instead of belly).
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 Wed Oct 21 2015 - 05:30:02 PDT
