[gmx-users] Request for PDB with the same label as in Gromocs topologies.

David Mobley dmobley at gmail.com
Tue Mar 28 17:29:57 CEST 2006


If I understand correctly what you're asking (that is, what atom types
in GROMACS correspond to what names in the PDB) this is really
nontrivial and there isn't one answer. In particular, each force field
has its own set of atom types. There are, for example, many different
types of carbon atoms that have different parameters depending on what
they are bonded to. Different force fields use different numbers of
atom types (which also have different names). Figuring out what atom
types you should use for a particular ligand isn't simply a matter of
figuring out which atom types correspond to which PDB names, as there
may not be a 1:1 correspondence (in particular, one PDB name may
become three different atom types in one force field, four in another,
and six in another). Then you have to somehow figure out what atom
type is the best for the molecule you are considering. This is one of
the particularly nasty parts of running calculations with ligands:
Coming up with parameters for the ligands is nontrivial.

Depending on what force field you are using, there may be automated
ways of doing this. I know a number of docking people who are using
the OPLS force field, and they must have some automated way of doing
"atom typing" for their ligands. (Atom typing is, if I understand your
question right, what you're talking about: Coming up with what atom
types best represent the atoms in the molecule of interest given a PDB
file). I don't know what utility people do for that. Another approach
is to use the AMBER force field (which the Pande group has ported to
GROMACS) and then develop ligand parameters using the ANTECHAMBER
package (which you can download online) with the GAFF (Generalized
Amber Force Field) for small molecules. These come out in AMBER
format, though (different units, etc) and need to  be converted to
GROMACS. A third approach is to use an online server that can generate
topologies for your ligand of interest. For example, prodrg can take
pdb/mol2/other files for ligands and generate certain
GROMACS-compatible topologies (but only for certain force fields).

I hope that gets at what you're trying to do. Basically the upshot is
that you need to decide what force field you want to use, and then
figure out how to do the atom typing appropriate for that force field.

An additional caveat is that you should make sure your ligand
parameters are compatible with the protein force field you are using
(that is, it would probably be a bad idea to use OPLS parameters for a
ligand while using AMBER parameters for a protein, or some such).

If you decide you want to use the AMBER force field, I can probably
give you a bit of help on generating GAFF parameters for your small
molecules; contact me off list. I don't have any experience with other
force fields at this point so you would be better off looking
elsewhere if  you want to use OPLS.

Incidentally, aside from getting force field parameters somewhere, you
may or may not need to also compute partial charges (I think in OPLS
partial charges are more or less linked to the force field, but this
isn't very much the case if you use GAFF or prodrg).


On 3/28/06, Mikko Hellgren <Mikko.Hellgren at mbb.ki.se> wrote:
> Hi all Gromacs users and developers, I wonder if anybody has pdb files
> for NAD+ (NADH in Gromacs), NADH (NADP) and Retinol (RTOL) with the same
> labels as in Gromacs topologies. This would save me a lot of time in my
> work since the atom names in my PDB for the ligands are different from
> Gromacs atom names, and it would take me a lot of time to figure out how
> to make them consistent to each other.
> I think it would be great if the developers of all the topologies in
> Gromacs also could submit the pdb file of the topology. This would save
> time (at least to me :-)) and help me to understand what all atom names
> in gromacs topologies stand for. I know that the problem of different
> atom labels is not due to Gromacs but it is the reason for the need of
> pdb:s for all molecules with topologies in Gromacs.
> If there is some other way to relabel atom names in an easy fashion, I
> would be very great full for any advice on that also.
> Best,
> Mikko Hellgren
> PhD student at Karolinska Institutet
> mikko.hellgren at ki.se <------- if you have the pdb I want, please send it
> to me.......
> _______________________________________________
> gmx-users mailing list    gmx-users at gromacs.org
> http://www.gromacs.org/mailman/listinfo/gmx-users
> Please don't post (un)subscribe requests to the list. Use the
> www interface or send it to gmx-users-request at gromacs.org.
> Can't post? Read http://www.gromacs.org/mailing_lists/users.php

More information about the gromacs.org_gmx-users mailing list