[gmx-developers] general force fields?

[Frank Wennmohs]

On Fri, Nov 23, 2001 at 04:32:36PM +0100, Frank Wennmohs wrote:
> On Fri, Nov 23, 2001 at 03:35:51PM +0100, David van der Spoel wrote:
> > On Fri, 23 Nov 2001, Frank Wennmohs wrote:
> > 
> > >What do you think about the inclusion of a more
> > >general force field in gromacs?
> > >Something like mmff9x which also applies to organic molecules.
> > >(topology generation would also be greatly reduced by this)
> > I would like that very much. What needs to be done I think is to separate
> > the forcefield from the topology, such that one can have a tool to
> > generate the topology (pdb2gmx, PRODRG, x2top) and then select a
> > forcefield afterwards. The force field should then contain tables for
> > atomtypes, charges etc. The most difficult thing is then the tables for
> > the bonded functions.
> > 
> > It would be very convenient to just have one topology and then select a
> > force field (e.g. GROMOS96/SPC, or AMBER/TIP3P) and compare systems
> > that are really identical, except for the forcefield.
> > 
> > To do this will require some coding, but most important, generation of a
> > tool that converts other force field to gromacs tables, or code to read
> > other force field files directly.
> 
> Yes, that's correct, as long as the definition of the forcefield does
> not differ that much from the gromacs/gromos forcefields.
> 
> The mmff9x is quite a different approach. 
> From the beginning laid out to cover a broader
> spectrum of compounds.
> It can deal with all chemical compounds by assigning atom types
> to the atoms. That's all. Everything else is derived from this.
> For this you got around 100(!) atom-types.
> And it's more expensive to calculate because of its functional form. 
> 
> For a detailed description have a look at
> Journal of Computational Chemistry, Vol. 17, Nos 5&6, 490- (1996)
> 

Sorry, this was somewhat misleading.
Of course the forcefield contains lots of other constants like
equ. distances, force constants and so on. But from the users
point of view only the atom types must be assigned.

Frank