[gmx-users] Combination rules for VdW parameters with CHARMM36 forcefield

[Shrinath Kumar]

Hi all,

I'm trying to simulate a protein with the CHARMM36 forcefield. I'm using
PME for VdW interactions and since the CHARMM forcefield
uses Lorentz-Berthelot combining rules I set *lj-pme-comb-rule
= Lorentz-Berthelot *in my mdp file. When I try run grompp with the
following command

gmx grompp -f npt_umbrella.mdp -c ang5.gro -p ../topol.top -n ../groups.ndx
-o npt5.tpr -po nptOut.mdp

I get the following warning

"WARNING 1 [file npt_umbrella.mdp]:
  You are using arithmetic-geometric combination rules in LJ-PME, but your
  non-bonded C6 parameters do not follow these rules."

So I checked my topol.top file to ensure that I was using the correct
forcefield and indeed topol.top contains *#include
"./charmm36.ff/forcefield.itp". *Looking at forcefield.itp the [ default ]
section is as follows.

; This GROMACS version was created from:
; CHARMM parameters: multiple files
; CHARMM topology: multiple files
; Please see forcefield.doc for files and references
[ defaults ]
; nbfunc        comb-rule       gen-pairs       fudgeLJ fudgeQQ
1       2       yes     1.0     1.0

>From sections 4.1.1 and 5.3.2 in the reference manual, comb-rule 2
corresponds to Lorentz-Berthelot rules. So I'm a bit confused by why the
warning is generated. I did also try with *lj-pme-comb-rule *set to
*Geometric*, but this again generated a note, instead of a warning this
time, that my non-bonded C6 parameters do not follow these rules.

Clearly, I'm missing something and I'd appreciate if someone could help me
figure out these warnings/notes.

I should also mention that I'm using Gromacs-2016-beta2 version (the latest
available on the cluster that I use) but I've also tried it on
gromacs-2016.4 on my local machine and got the same warnings.

Regards,
Shrinath