GROMACS 2026.2 duplicate type-9 dihedral errors with CHARMM36-feb2026 and CGenFF 5.0

User discussions
1

Hi everyone,

I am performing a protein–ligand MD simulation using

GROMACS 2026.2
CHARMM36-feb2026
CGenFF 5.0

Ligand topology was generated using the latest cgenff_charmm2gmx.py
(March 2025 David Moody version).

Command:

python3 cgenff_charmm2gmx.py DL DL_fix.mol2 DL.str charmm36-feb2026.ff

The conversion completed successfully.

However, during

gmx grompp

I obtain multiple errors such as

Encountered a second block of parameters for dihedral type 9

Example:

CG321 CG311 CG311 NG3C51

CG331 CG311 CG311 NG311

I verified that these dihedral definitions already exist in

charmm36-feb2026.ff/ffbonded.itp

and are also present in the generated dl.prm.

I also receive

Invalid directive impropertypes

for the impropertypes section in dl.prm.

Questions:

  1. Is this a known compatibility issue with GROMACS 2026.2?

  2. Is there an updated version of cgenff_charmm2gmx.py?

  3. Should duplicate dihedral blocks be removed from dl.prm, or is another workflow recommended?

This work is for a research publication, so I would like to follow the officially recommended workflow.

Thank you.

grompp_output.txt (9.1 KB)

topol.top (1.4 KB)

2

If the dihedral is already in ffbonded.itp, then it should not be in the ligand .prm file. If it’s there, you used an older version of CGenFF to generate the ligand topology than the current distribution. Any duplicate should be deleted from the .prm file; it has nothing to do with the conversion. You should double-check by regenerating the ligand topology and ensuring that CGenFF 5.0 is being used on the CGenFF server (we are in a transition period, for a time 4.6 and 5.0 were options, but 5.0 should be the default).