Dihedral bonds disappearing/ignored during GROMPP step

User discussions
1

GROMACS version: 2021.4-Ubuntu-2021.4-2
GROMACS modification: No

I’m experimenting/playing with gmx, and found that some bonds from my topology file is not present in the .tpr file.

From topology file:
[ dihedrals ]
(…)
** 18 5 7 10 9 **
** 1 5 18 19 9 **
** 6 5 18 19 9**

from outputtpr.txt:
Proper Dih.:
** nr: 175**
** iatoms:**
(…)
17 type=110 (PDIHS) 17 4 6 9
18 type=111 (PDIHS) 6 4 17 18

First off the “nr: 175” is completely off, there is 34 dihedrals.
But importantly, the "1 5 18 19 9 " is not in the output file. The atomtypes are NH3-CT1-C-NH1, which is represented in the CHARMM27 forcefield i use:
NH3 CT1 C NH1 9 0.00 2.5104 1

I do get some warnings related to charge when using gmx grompp, but nothing relevant to dihedrals. I really confused as to how to troubleshoot and fix this, any ideas?

2

It’s difficult to say much without your whole topology file. Is there only one molecule (one type and one copy of it) in your system?

3

topol.top (5.9 KB)
tpr_content.txt (162.3 KB)

Thank you for replying.
Yes there is only 1 molecule, it is a Methionine AA extracted from the 6lzm molecule.
I’ve attached the files i believe to be relevant.

4

conf.gro.txt (935 Bytes)

5

Thanks. Can you generate the tpr with a supported version of GROMACS? 2023/2024?

I agree that the number 175 looks off, but it is not the number of dihedrals, but the number of values stored, i.e. 5 per dihedral and there are 35 dihedrals. Currently, I can’t explain why the 1 5 18 19 dihedral is missing.

6

i.e. 5 per dihedral and there are 35 dihedrals
Ah okay thanks, that makes sense then.

I have now switched to Gromacs 2024.2, following the Quick-And-Dirty installguide. All regression-tests passed.

I have regenerated both the topology and grompp dump with this version.
gromppout.txt (161.7 KB)
topol.top (5.7 KB)

My steps to get here:
gmx pdb2gmx -f 6lzm.pdb Selecting charmm27, then SPC
Manually remove all but the first residue (MET), and all bonds including atoms not in this residue, from .top and .conf file
gmx grompp -f sim.mdp
gmx dump -s topol.tpr > gromppout.txt

Another detail i noticed, don’t know if it is relevant.
This dihedral from the .top: 7 10 13 14 9
Is present twice in the .tpr
28 type=114 (PDIHS) 6 9 12 13
29 type=115 (PDIHS) 6 9 12 13

7

Could you post the tpr file, please? I can’t generate it without the mdp file.

8

topol.tpr.txt (63.6 KB)
The site wont let me upload .tpr directly, so i just manually added a temporary .txt extension

Also the .mdp file i used in grompp, for good measure:
sim.mdp (63 Bytes)

9

Thanks. I haven’t found a good explanation (yet?). @hess, do you have any idea why some of the dihedrals in the topology don’t show up when dumping a tpr?

10

The tpr is correct. The missing dihedrals has all parameters zero, so it is removed by grompp. Another dihedral gets duplicated as there are two sets of parameters for this type 9 dihedral.

11

Another dihedral gets duplicated as there are two sets of parameters

I did not know about multiple sets of parameters, thanks for clearing that up.

The missing dihedrals has all parameters zero

Ah yes you’re right, i had mixed up which atomtypes were in the dihedral. There is no NH1 in this example… sorry for wasting your time.

It add’s up now, thanks for the help!

Best, Daniel