Re: Protein-ligand simulation

From: Kenno Vanommeslaeghe (kvanomme_at_rx.umaryland.edu)
Date: Tue Jun 25 2013 - 13:09:27 CDT

Hi James,

On 06/25/2013 01:58 AM, James Starlight wrote:
> Today I've used CGenFF at paramchem.org <http://paramchem.org/> for my
> test molecule (negatively charged cyclic guanosine mono phosphate).

Actually, I forgot to mention CGenFF is not meant to be used for common
biomolecules that are covered by the more specialized CHARMM force fields.
cGMP is already in the CHARMM nucleic acid force field ( PRES CY35 ), so
there's no force field assignment or optimization required whatsoever; all
you need is a proper psf file. I could tell you how to generate it with
the CHARMM program, but I'm not familiar with psfgen; the other denizens
of this mailinglist may be able to help you out with that.

> How should I convert CGenFF output (str file) to the Namd par format ?

In your case, it's better to use the nucleic acid force field, but just
for future reference, NAMD actually uses the CHARMM parameter format
(prm). Our str file is essentially a concatenated CHARMM topology (rtf)
and prm file with some lines of "glue" around them. Both the rtf and prm
sections start with a few lines starting with "*", and end with "END", so
the "conversion" consists of copy-pasting the correct part of the str file
to a separate file. Also note that an str file can be fed directly to NAMD
- it will cleverly ignore everything but the parameter section.
http://mackerell.umaryland.edu/~kenno/cgenff/program.html#namd

> RESI tmp2.pdb -1.000 ! param penalty= 196.000 ; charge penalty= 111.390
> GROUP ! CHARGE CH_PENALTY

It looks like in the case of this particular molecule, the penalties are
woefully pessimistic. The penalties are by definition estimates, and we
built in a little bit of systematic overestimation to err on the safe
side. Because of this, one very occasionally encounters cases at the other
tail of the bell curve where the penalties are unrealistically high, and
this appears to be one of them. But again, using the nucleic acid force
field for this one would still be preferable.

Cheers,

        Kenno.

This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:21:21 CST