[gmx-users] CHARMM force field implementation in Gromacs : retrieval of bonded/non-bonded parameters

[Mark Abraham]

> Hello,
>
> I'm trying to use an implementation of the CHARMM force fields in Gromacs
> and I'm still a little bit confused about how Gromacs retrieves the
> bonded/non-bonded parameters when the .top file is created.   Concretely,
> I have :
> 1. a ffcharmmnb.itp where I have defined specific 1-4 LJ parameters in the
> [pairtypes] section
>
> 2. a ffcharmmbon.itp where I have defined bonded parameters
>
> 3. a ffcharmm.rtp where I have defined the topology of amimo acids, etc.
> Since the [bonds], [angles], [dihedral] sections are optional in the .rtp
> file, I haven't defined them.
>
> How all these parameters are taken into account when I generate the .top
> file with pdb2gmx?

Section 5.5.1 of the manual describes the behaviour of pdb2gmx in
interpreting the contentes of the .rtp file in conjunction with the .itp
files.

> When I look in the .top file, I see an [atoms] section
> but no [bonds] or [pairs] etc.

In particular, you will read there that angles are automatically
generated. Perhaps that should be amened to include the proviso that they
are automatically generated only between pairs of bonds. Since you have
defined no bonds, you will get nothing else.

> and the charmm ff doesn't seem to be
> included with an #include command. So, are all my specific 1-4 parameters
> really taken into account in this case? In fact, I have tried to add somd
> bonds and improper but that generates a chunk of warnings when I use
> grompp.

Go back and try adding bonds to your .rtp and see how you go.

Cheers,

Mark