[gmx-users] force field coding

Erik Lindahl lindahl at csb.stanford.edu
Mon Oct 20 06:40:01 CEST 2003

Hi Corina,

There are a couple of steps:

1. Find out exactly what the interaction forms are for bonds, angles, 
proper dihedrals (sometimes called torsions), and improper or 
out-of-plane torsions/dihedrals. If we can represent these, we're set,  
but in some cases we have to take a factor 2 into account. If not, it 
depends on the type of interaction you need. For instance: using a new 
type for bonded interactions isn't too difficult (see e.g. how we use 
both standard and Gromos96 bonds), but having nonbonded interactions 
between triplets of atoms would be very difficult to do...

2. Check how the force fields use atom types. Does each atom type 
correspond to unique bonded and nonbonded parameters (like Gromos96 & 
Amber), or is it like OPLS, where there are more nonbonded than bonded 

3. Implement the atom types in ffXXX.atp (easy), and add the nonbonded 
parameters to ffXXXnb.itp. Most forcefields use combination rules and a 
small number of atom types, but Gromacs can handle different parameters 
for every single pair of atoms. You will need to check whether your 
force field uses  c6/c12 or sigma/epsilon parameters. Some 
implementations don't specify sigma, but the minimum radius - that's 
easy to convert, though.

4. Implement the parameters for bonds, angles and dihedrals in 
ffXXXbon.itp. This can be complicated if your forcefield has thousands 
of atom types like OPLS, but in most cases it's easy.

5. At this point you have a finished force field. To build topologies 
automatically you also want to implement the building blocks in 
ffXXX.rtp, but don't write it from scratch - just pick one of the 
existing force fields and edit that.

On Sunday, October 19, 2003, at 02:54 PM, CORINA BARBU wrote:

>   Hello!
> I need to use a different force field (MM2/MM3) from those already 
> implemented
> in GROMACS. I wonder how difficult is to implement it. At the first 
> sight it
> seems to me that I need to provide an *.itp file (ff--bon.itp, 
> ff--nb.itp,
> ff--.atp) for my case, with parameters taken from literature. I would 
> also have
> to indicate what terms to consider in the expresion of the energy 
> functional.
> And this can be indicated in the column 'func' in the *.itp file 
> containing the
> definition of the molecule. If GROMACS already has the 
> interaction-functions
> that I need, then I just need to choose some of them. If not, I have 
> to code
> what doesn't exist.
> I'm interested in the coding part...what would I need to code for a 
> new force
> field?
> Thanks very much! Corina
> _______________________________________________
> 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.

More information about the gromacs.org_gmx-users mailing list