Re: [Towhee-users] Geometry of large organic molecules in bulk phase

[Marcus M. <mar...@us...>]

On Wed, Apr 29, 2009 at 3:28 PM, Tandia, Adama <Ta...@co...> wrote:

> I would like to compute the optimized geometry of a large organic
> molecule in the bulk phase. I would like to use an initial box with
> randomly distributed molecules and use a CBMC. My molecule has two
> torsion angles. Can anyone who had done this type of calculation before
> help me with initial input file and force field file that I can started
> with.

If it only has 2 torsion angles then it is not overly large and it
should be fairly easy to simulate using the default CBMC algorithms.

The OPLS_Propanamide example is a pretty good place to start.  I would
make a handful of changes to that towhee_input file for your purposes.

0) The obvious - input your molecule and make sure the nonbonded
potential and coulombic potential are correct for whichever forcefield
you are using.
1) I would remove the cbmc_formulation and cbmc_setting_style
variables and just use the defaults.
2) A real production run is going to need at least 10000 cycles and
probably a few equilibration runs of the same length.
3) "Optimized Geometry" is a concept from the 0 K world of quantum
mechanics.  In a Monte Carlo simulation you will instead end up with a
distribution of bond lengths, bending angles, and torsions.  You will
need to sample the configurations by setting the moviefreq variable to
output towhee_movie files.  I find that somewhere between 100 and 1000
snapshots is about right to get smooth distributions so I would
suggest setting moveifreq in the range of
nstep/1000 to nstep/100.  After the simulation completes you can use
the analyse_movie utility to calculate the distribution of all of the
intramolecular terms of your molecule.

Marcus

-- 
Marcus G. Martin
Director, Useful Bias Incorporated
88 Martinez Road
Edgewood NM 87015-8222
ph. (505) 286-4457
www.usefulbias.com
www.photobirder.com