[gmx-users] how to get pdb2gmx to use an arbitrarily located forcefield.ff directory?

[Christopher Neale]

Thank you, Francesco, for the assistance. However, I still can not get it to work (see below). Can you please specify exactly what you have done and perhaps show a listing of the directory structure?

My command was: pdb2gmx -f protein.pdb -ff charmm36

I tried this in a number of ways, all eventually failed:

neither GMXLIB or GMXDATA set:
Error: "Could not find force field 'charmm36' in current directory, install tree or GMXDATA path."

# define GMXDATA
export GMXLIB=""
export GMXDATA=~/gromacs/CHARMM36/share 
Error: "Could not find force field 'charmm36' in current directory, install tree or GMXDATA path."

# define GMXLIB:
export GMXLIB=~/gromacs/CHARMM36/share/gromacs/top
export GMXDATA=""
the charmm36 ff is found, but after selecting a water model, I get:
Error: "Library file residuetypes.dat not found in current dir nor in your GMXLIB path."

# define both:
export GMXDATA=~/gromacs/CHARMM36/share  
export GMXLIB=~/gromacs/CHARMM36/share/gromacs/top 
the charmm36 ff is found, but after selecting a water model, I get:
Error: "Library file residuetypes.dat not found in current dir nor in your GMXLIB path."

# set GMXLIB to the regular installation:
export GMXDATA=~/gromacs/CHARMM36/share  
export GMXLIB=/usr/local/gromacs-4.5.5/share/gromacs/top/
Error: "Could not find force field 'charmm36' in current directory, install tree or GMXDATA path."

# the reverse of the idea above:
export GMXDATA=/usr/local/gromacs-4.5.5/share
export GMXLIB=~/gromacs/CHARMM36/share/gromacs/top
Error: "Library file residuetypes.dat not found in current dir nor in your GMXLIB path."

Here is a listing of the artificial directory structure that I created to try to get this to work:

[nealec at pegasus check]$ ls ~/gromacs/CHARMM36/
share
[nealec at pegasus check]$ ls ~/gromacs/CHARMM36/share/
gromacs
[nealec at pegasus check]$ ls ~/gromacs/CHARMM36/share/gromacs/
top
[nealec at pegasus check]$ ls ~/gromacs/CHARMM36/share/gromacs/top/
charmm36.ff
[nealec at pegasus check]$ ls ~/gromacs/CHARMM36/share/gromacs/top/charmm36.ff/
aminoacids.arn    aminoacids.n.tdb  aminoacids.vsd  dna.arn    dna.n.tdb     ffnabonded.itp     forcefield.doc  ions.itp    popc.itp   rna.hdb    rna.rtp   tip3p.itp   watermodels.dat
aminoacids.c.tdb  aminoacids.r2b    atomtypes.atp   dna.c.tdb  dna.rtp       ffnanonbonded.itp  forcefield.itp  lipids.hdb  rna.arn    rna.n.tdb  spce.itp  tip4p.itp
aminoacids.hdb    aminoacids.rtp    cmap.itp        dna.hdb    ffbonded.itp  ffnonbonded.itp    gb.itp          lipids.rtp  rna.c.tdb  rna.r2b    spc.itp   tips3p.itp

Thank you,
Chris.

-- original message --

Hi Chris,
I don't know if thngs changed in gromacs 4.6.x, but I succesfully did what
you are triying to do setting the two variable GMXLIB and GMXDATA at the
same time.

Francesco


2013/3/5 Christopher Neale <[hidden email]>

> Hello,
>
> I have downloaded the charmm36.ff directory and would like to use it with
> pdb2gmx. Everything works fine if I put it in the current directory or the
> share/gromacs/top directory of the binary that I am using. However, I'd
> like to be able to put the charmm36.ff directory in an arbitrary place.
>
> I tried setting GMXLIB, no luck. I also tried setting GMXDATA and also,
> separately, creating a directory structure share/gromacs/top/charm36ff and
> then setting GMXDATA to share/ (which seems absurdly convoluted, but is
> implied that it should work by the description of GMXDATA here
> http://www.gromacs.org/Documentation/Terminology/Environment_Variablesand the text at the output of pdb2gmx:
>
> Fatal error:
> Could not find force field 'charmm36' in current directory, install tree
> or GMXDATA path.
> For more information and tips for troubleshooting, please check the GROMACS
> website at http://www.gromacs.org/Documentation/Errors
>
> I can get around this by making a link in the current directory and then
> using sed to change the directory structure in the resulting .top file,
> after which I can remove the link in the .top file.
>
> Thank you for any advice,
> Chris.