[gmx-users] newbie question about GROMACS input files

[Mark Abraham]

On 9/11/2011 10:08 AM, Markus K. Dahlgren wrote:
> Hi all!
>
> I am new to the Gromacs mailing list and just started looking into Gromacs for
> MD simulations. I have been working on converting BOSS output into automatic
> NAMD input for small molecule ligands and now want to create a similar program
> for Gromacs input. I have a question regarding the input files, as I find the
> large number of files confusing.
>
> What is the required files to start a Gromacs simulation? Using pdb2gmx you need
> a pdb file. To get the parameters, ffbonded.itp, ffnonbonded.itp, atomtypes.atp
> and *.rtp are needed?

To generate a topology pdb2gmx requires .atp and .rtp files, and maybe 
some other stuff depending on what functionality is required. To use a 
topology, grompp requires that the #include statements generated by 
pdb2gmx make sense, so the .itp files are only then required. Updating 
residuetypes.dat can be wise.

>   Is that all the files needed if these files completely
> describe the input molecules? Also, does the new parameters for the ligand need
> to go into the existing parameter files inside the XXX.ff folder? Or how can new
> *.itp, *.rtp and *atp files be defined and called upon by pdb2gmx?

You can make a new local copy of the xxx.ff folder and modify that. 
However, it is more work to create an .rtp entry for a small molecule 
and use pdb2gmx than it is to write a molecule.itp file directly (see 
example in Chapter 5 of the manual). Such an .itp file can probably have 
[nonbond_params] or [atomtypes] fields in it, or implicitly refer to a 
modified ffnonbonded.itp.

>
> I also read about the topology file in the manual. If a topology file would be
> written by the program, what other files are needed? The topology would have
> all atomtypes, names, charges, bonds, angles, and dihedrals. Would it be enough
> to supply a nonbonded.itp file together with the *.top file and the pdb file?

Yes. Note that parameters for bonded interactions can be provided 
directly in the .itp where the interaction is defined, rather than by 
reference to a bonded interaction type entry.

>
> Finally I have a question about the *.gro files. Are they always needed when
> starting a simulation? For a new simulation, the last three columns (vx, vy,
> vz) can be left out?

A coordinate file is required, but its format can be to suit your 
convenience. Particularly if you are generating velocities, the velocity 
values can be irrelevant.

>
> Sorry about asking about such trivial matters, but the number of files was
> confusing and I was not sure what files were needed if I would bypass the
> pdb2gmx step and write the topology file directly.

You can.

Mark

>
> Thanks in advance for any insight you can offer.
>
> Best,
> Markus
>
>
>