[gmx-users] Overriding Wildcard dihedraltypes

[hubert santuz]

Dear Gromacs Users,

I'm trying to set up a molecule into the charmm36 force field for lipids.
I correctly obtained a good itp file from a charmm file.
In this itp file, a wildcard dihedraltype (from the ffbonded.itp) is 
overrided by an explicit dihedraltype :

In ffbonded.itp :
X    CTL1    CTL2    X    9    0.00    0.8368    3
In my itp file:
CRL1   CTL1   CTL2   CTL2     9    180.00       0.50961    3
CRL1   CTL1   CTL2   CTL2     9    180.00       0.91002    2
CRL1   CTL1   CTL2   CTL2     9    180.00       1.00332    1

In my topol.top:
#include "charmm36.ff/forcefield.itp"
#include "my_molecule.itp"


Unfortunately, this override is not taking in account by gromacs.
When I dump the tpr, I do not see this 3 types in the "functype" lines.
However, when I add this override into the ffbonded.itp file before the 
wildcard dihedraltype like this :

In ffbonded.itp:
CRL1   CTL1   CTL2   CTL2     9    180.00       0.50961    3
CRL1   CTL1   CTL2   CTL2     9    180.00       0.91002    2
CRL1   CTL1   CTL2   CTL2     9    180.00       1.00332    1
X    CTL1    CTL2    X    9    0.00    0.8368    3

In this case, I can see clearly the 3 dihedraltypes in my tpr.

More interestingly, when I try to override an explicit dihedral type 
from ffbonded.itp (like CEL1 CTL2 CEL1 HEL1) into my itp file, I get a 
warning from grompp and I see the override in the tpr file.
It seems that wildcard dihedral types cannot be override (at least with 
the function 9, I didn't test with other forcefields).

For the moment, the only solution that I have found is either to :
    - add the override in the ffbonded.itp which is not a good idea.
    - or add 2 new atoms with new names but with the same parameters 
than (CTL1 and CTL2) in my itp file and change the dihedraltype 
according to these two new atoms.

Without this override, I'm not able to reproduce the correct dihedral 
energy from a SPE calculation between gromacs and namd... (but with the 
override, I am!)

Have people already experienced this case ?
Also, I can provide test files (itp, top, pdb) with only 4 atoms and 
this dihedral.

Thanks for the help!
Best regards,

Hubert