OK, so reading the Gromacs manual (which is a really good how-to on many
things MD, I might add), I see Sphere (g = 1) and Cylinder (g = 2), so from
the rest of your message I am assuming you meant "option (g=1 sphere)."
That's something that Amber can do, but if your system is up to 10,000
atoms then depending on the type of cutoffs you want (a raw Coulomb
interaction will run into lots of problems) you may be better off going to
periodic boundary conditions already. Vacuum simulations are NOT
absolutely faster than simulations in water. The problem is that vacuum
simulations often require a big cutoff to the tune of 15 or 18 Angstroms,
compared to condensed-phase simulations with PBC which take short cutoffs
and the O(N log N) P3M interactions (also called PME).
Another thing to note is that, if all you want is for the molecules to be
confined to a particular region, periodic boundary conditions with NVT
(fixed number of particles, volume, and temperature) or NVE (canonical
ensemble, fixed particle count, volume, and energy) will also give you this
effect. If a particle tries to wander out one side of the box, it just
checks back in the other and sees the same neighbors it used to, perhaps
from a different side. Welcome to the Hotel Periodic.
So if you just want to see molecules bumping into each other in a confined
space I would recommend just taking NVT in PBC, with PMEMD. No spherical
restraint force necessary--the paper you cite sounds like it's using a
duct-tape way to emulate stochastic boundary conditions, but the artifacts
you get are not necessarily any more benign than what would happen with
PBCs. If you just want to put in a number of copies of each molecule and
watch the structures they form, make your boxes cubic, create some PDB
structures of your INDIVIDUAL molecules, and use my AddToBox utility to
randomly place multiple copies of them around a box of your liking. This
will give you a PDB file with an initial structure. For vacuum
simulations, use the tleap command "setBox x vdw 10.0" (where x is the name
of the structure you read in, as in "x = loadPdb <name of PDB>"). For
solution phase simulations, use AddToBox again to add water in the
proportion you want (tleap's solvateBox will not necessarily place the
amount of water you're looking for if there is a target molality). In all
cases, use ChBox to edit the final box dimensions of your system back to
whatever you were using in with AddToBox (tleap will make its own judgments
about what the box dimensions should be, but AddToBox knows about PBCs
already so you can reset the box dimensions).
When you simulate, for the vacuum case and optimal speed I'd recommend
bumping up the cutoff and decreasing the grid density proportionately. If
you're happy enough with the accuracy of the default settings, bump the
cutoff up to 12 and select a value of nfft1, nfft2, and nfft3 (remember,
cubic box) that gives you a grid spacing just shy of 1.5A. That will cut
down on the reciprocal space work by up to three times depending on how
much empty space is in your simulations. For condensed phase simulations,
just stick to the default run conditions (these will be 8A cutoff, grid
spacing just shy of 1A).
Dave
On Mon, Apr 4, 2016 at 2:52 PM, Kenneth McGuinness <
kenneth.mcguinness.gmail.com> wrote:
> Hey Dave - great to hear from you!
>
> I wrote my comments in-line to help with following the conversation
>
>
>
> On Mon, Apr 4, 2016 at 2:32 PM, David Cerutti <dscerutti.gmail.com> wrote:
>
> > Hi Ken!
> >
> > The best advice I can give here is that Amber isn't a program that does
> > hard-sphere potentials, so if you want particles to suddenly hit an
> > infinite wall when they travel within (or outside of) some distance of
> one
> > another you'll need a separate code.
>
>
> The type of potential I am looking to implement is the one implemented in
> section 4.3.2 with option (g=2 sphere) in this link:
>
> ftp://ftp.gromacs.org/pub/manual/manual-5.0.7.pdf
>
> I tried gromacs but it does not implement GPU usage yet with vacuum/and no
> pbc -speed is the biggest need right now- Amber does conveniently
>
>
>
> > The closest you could get would be to
> > use the nmropt (NMR optimization) features to implement HARMONIC
> restraints
> > on either side of a flat-bottom potential.
>
>
> Not sure how to do this in a 3D sense (spherical) in terms of the DISANG
> options and restraint files (gromacs has a simple format) I have not
> wrapped my head around Amber's way of doing it yet :-)
>
> Particles connected by such a
> > potential experience no forces (at least, no forces due to one another)
> so
> > long as they are within some small distance of on another, but begin to
> > experience harmonic attraction as the distance crosses that threshold.
> >
> > Depending on what exactly you want to simulate, I can recommend different
> > programs. How many particles (atoms) will your system have?
>
>
> ~1000-10,000 atoms
>
>
> > Do you want
> > periodic boundary conditions, to simulate the effect of having a fairly
> > elaborate density of material, or are you content to see just a few of
> > these molecules interact in the gas phase, off in an otherwise empty
> region
> > of space?
> >
>
> I am setting up a screening process for molecular interactions
>
> 1. Speed is the most important part so I intend on using the gas phase
> without periodic boundaries
> 2. Once I see interactions worth pursuing in (1) I will then add more
> molecular details (implicit, explicit solvent, pbc etc.) in the force field
> to rule out interactions worth testing in the lab.
> 3. The spherical potentials prevent the molecules from drifting too far
> into space because of the lack of viscosity
>
> I am really grateful for your time!
>
> Any more details I can give? Please lmk
>
> Thank you Dave!
>
>
>
>
> >
> > Dave
> >
> >
> > On Mon, Apr 4, 2016 at 2:03 PM, Kenneth McGuinness <
> > kenneth.mcguinness.gmail.com> wrote:
> >
> > > Hi everyone,
> > >
> > > This is my first post to this forum and I am grateful to be a part of
> it!
> > >
> > > I am relatively new to Amber as well. Sorry if this post goes over
> > material
> > > that someone else has posted before. I have looked as far as I could
> and
> > I
> > > did not find an exact match on what I am looking to do.
> > >
> > > So without going into the entire project,
> > >
> > > I would like to run simulations in vacuo using spherical flatwell
> > > potentials.
> > >
> > > I am not sure how to implement this - however this article has
> > implemented
> > > the method (this paper is similar to my work - and the method will
> > greatly
> > > help my work)
> > >
> > > http://pubs.acs.org/doi/abs/10.1021/jp501503x
> > >
> > > Any help in setting this up would be very much appreciated - I am still
> > > trying to decipher what section 24.1 means in the Amber doc. Please let
> > me
> > > know any details that are needed from me in order to progress.
> > >
> > > I greatly appreciate any and all help!
> > >
> > > Thank you!
> > >
> > > With gratitude,
> > >
> > > ~Kenneth
> > >
> > > Kenneth N. McGuinness Ph.D
> > > Research Associate
> > > Advanced Science Research Center CUNY
> > > Adviser: Dr. Rein Ulijn <http://www.ulijnlab.com/>
> > > Cell: 928-925-7693
> > > _______________________________________________
> > > AMBER mailing list
> > > AMBER.ambermd.org
> > > http://lists.ambermd.org/mailman/listinfo/amber
> > >
> > _______________________________________________
> > AMBER mailing list
> > AMBER.ambermd.org
> > http://lists.ambermd.org/mailman/listinfo/amber
> >
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Mon Apr 04 2016 - 13:00:03 PDT
