[gmx-users] regarding SwissParam topology

[Justin Lemkul]

On 6/10/15 9:22 AM, soumadwip ghosh wrote:
> Thanks Justin,
>                      actually I did look into what you suggested. I
> made a carbon nanotube topology with the help of pbd2gmx. Now, the
> atomtypes specified in the .rtp file was CA as it is aromatic carbon
> atoms. Then I looked into the cnt.itp file and the kb, force constant
> and other terms in it were missing and proceeding with it made grompp
> complain about 'missing parameters in the itp file'. Hence I looked

Because they're not missing.  If values are not in the .itp or .top explicitly, 
they are referenced from ffbonded.itp.  If grompp can't find them there, you get 
an error about missing parameters.  This is how every force field works in 
GROMACS, except for the GROMOS force fields, which are oddballs and use #define 
macros.

> into ffbonded and ffnonbonded.itp files and manually filled the
> missing columns in the cnt.itp. It was a trivial job as the cnt was
> constituted with similar carbon atoms and terminal hydrogens. The
> interaction of this nanotube with a sigle stranded DNA was as
> expected. There was helical wrapping and non-bonding interactions
> could be captured perfectly using this model. However, for a similar
> system like fullerene interactions are not arising with DNA when an
> alternative SwissParam topology is used.
>

So what are the differences in the nonbonded parameters here?  You're getting 
sidetracked on bonded parameters, which I would bet strongly are totally 
irrelevant here.

The structure of a CNT and a fullerene are different.  The curvature is 
different and the physical properties may be different.  Maybe it takes longer 
for an interaction to arise, or maybe the CNT worked out by luck and the 
fullerene is revealing deficiencies in the physical model (force field).

> I am thinking of changing those atomtypes from CB to CA in
> fullerene.itp file and have a trial whether the interaction patterns

How different are the parameters for CB vs. CA?

> are changing or not. I am quite sure if I generate the topology by
> pdb2gmx and then fill up the missing parts from the forcefield files,
> the desired interaction would arise because it happened for the cnt

"Filling up" bonded parameters, again, is irrelevant.  You should get the same 
bonded parameters either way.

> case and anyways making an rtp and topology for a small molecule like
> fullerene wont be an issue. I was just wondering why this .itp from
> SwissParam behaves so weirdly.
>

You're comparing applies and oranges.  One system (CNT) works as expected, one 
(fullerene) does not.  Unless you're getting non-reproducibility among the same 
system, you're not looking at the right things.

-Justin

-- 
==================================================

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================