Segmentation fault on mdrun

User discussions
1

GROMACS version: 2023.2
GROMACS modification: Yes/No

Hi,

  1. I’m wondering if ‘LINCS WARNING’ is showing up as ‘segmentation fault’ in version 2023.2. My guess is that in version 22.4, I could see a similar error with LINCS WARNING message, but since using version 23.2, I haven’t seen the phrase LINCS WARNING in the err log.

  2. I’m simulating a polyphenol molecule. The .itp file was newly generated by ATB (https://atb.uq.edu.au/index.py).

I added water and ions to the ligand without proteins. I used gromos54a7 and spc. I can’t change ff because the topology builder (ATB) site limits ff to gromos54a7.

I keep getting the error message below in the MD step after the EM, NPT, NVT steps. The simulation fails to complete and aborts.

starting mdrun ‘ALL0 in water’
5530000 steps, 11060.0 ps (continuing from step 530000, 1060.0 ps).
train.sh: line 21: 1475476 Segmentation fault (core dumped) gmx mdrun -deffnm md1

I attach the em, npt, nvt, md.mdp files I used.

Thanks.

em.mdp (1.2 KB)
md.mdp (2.2 KB)
npt.mdp (2.2 KB)
nvt.mdp (2.2 KB)

2

This is almost certainly the system that is exploding. I would suggest the following to see if it helps:

In md.mdp:
The reason why I think your system crashes is because of oscillations caused by your low tau-p with a Parrinello-Rahman barostat. Switch to c-rescale instead. c-rescale can be used for both equilibration and production.
See also my suggestions for npt.mdp below.

I addition I would recommend the following changes, especially if the above did not help.

In em.mdp:
Turn on the position restraints for protein and ligand (define = -DPOSRES -DPOSRES_LIG)
Increase nsteps to 25,000 and check the log output.

In nvt.mdp:
Also here I would suggest using position restraint for protein and ligand.
Increase nsteps to 100,000.
Set constraints = hbond.
I would also argue that you don’t need annealing to heat the system.
I think your tau-t is unnecessarily low. I think you would do fine with 1ps.

In npt.mdp:
Here I also suggest a higher tau-t and to only constrain the hydrogen bonds.

I hope that helps.

1 Like