[gmx-users] Re: DNA-protein

Michael Brunsteiner mbx0009 at yahoo.com
Thu Mar 3 16:16:13 CET 2005

Anton Feenstra wrote:

> > Dear All,
> > 
> > please forgive me if it sounds ignorant.
> >  
> > I'm just wondering whether gromacs is also very reliable to simulate the 
> > DNA-protein interactions, for example by BD or MD. It will be very
> > helpful if some publication available for my reference.
> There must be plenty of papers (publications) on DNA simulations in the 
> OPLS forcefield (which is available in Gromacs), and there are a few 
> recent papers by Van Gunsteren and Oostenbrink on DNA simulations in the 
> (new) Gromos g53a6 forcefield, which, on request, is also available in 
> Gromacs.

What you say might be true, though it is somewhat misleading.
As has been said many times before in this forum, DNA
with Gromos, as implemented in Gromcas does NOT work.
The order of the atoms in the [atom] entries in the rtp
files do not correspond to the order commonly found
in DNA pdb-files, the *.hdb files do not include entries for
DNA and NO terminal DNA residues are defined. Of course you
can go there and do the required additions and modifications
by hand, but this is a major task and, in particular when
it comes to the charges on the terminal residues, some, in my opinion
rather dangerous guess-work is required. This is true for both g53a6 
and g43a*.
As for opls: DNA is NOT included in the opls FF files that
come with Gromacs (not for version 3.2.1 anyway).
Apparently a remedy can be found in: 
but this project seems to be in an early stage, not 
thoroughly tested, and thus not advisable for newbies.

I believe it would be a good idea of the developers
to clearly state these facts somewhere on the Gromacs
webpage, since ever so often I see mails of people 
who do have problems simpulating DNA in the users mailing list.


Celebrate Yahoo!'s 10th Birthday! 
Yahoo! Netrospective: 100 Moments of the Web 

More information about the gromacs.org_gmx-users mailing list