Segmentation fault -- no error message

GROMACS version: 2021.4
GROMACS modification: /No

Hi all,

I’m trying to simulate a myeloperoxidase protein monomer in water at ambient conditions, but I’m continuing to run into an problem that I don’t know how to solve. I’ve taken the crystal structure (3F9P) from RCSB and parameterized the metal-containing center using MCPB.py, but I’m unable to energy minimize my structure. Specifically, GROMACS returns the following output when I use mdrun for an energy minimization (steepest descent, dt = 1 fs):

Steepest Descents:
Tolerance (Fmax) = 1.00000e+03
Number of steps = 1000
/var/spool/torque/mom_priv/jobs/637814.trcis002.hpc.wvu.edu.SC: line 36: 11726 Segmentation fault (core dumped) gmx mdrun -deffnm minim -ntmpi 1 -ntomp 1

At this point, the energy minimization does not run (the output trajectory file is empty), and there is no fatal error returned from the grompp command. I would expect to get LINCS warnings at the first time step if this issue arose from an unstable starting configuration, but I have visualized my input structure file anyway without finding any obvious problems. The size of my simulation is reasonable (46,000 atoms). I’ve also attempted running the energy minimization using several different versions of GROMACS (2016.6, 5.1.5, and 2021.4), but the issue persists.

I would appreciate any insight into what could be causing this problem, especially because I’m not getting an error message from GROMACS (so could be machine related). I can provide any additional information (pdb file, mdp options, topology files, gmx command line inputs) as needed.

Best,
Jordan Chapman
Graduate research assistant
West Virginia University

If you can provide such problematic tpr file and/or the .top/.gro and command line you used to generate it, it would help diagnosing the problem.

Sure thing. I’ve included a link below containing the input structure file, molecular and system topology files, the tpr file generated, and the MD options specified for the energy minimization. The command lines I used are:

gmx grompp -f minim.mdp -c initial_config.gro -r initial_config.gro -p topol.top -o minim.tpr
gmx mdrun -deffnm minim -ntmpi 1 -ntomp 1

[https://drive.google.com/drive/folders/1A_8Ck6ya59ai0BNPpPYVYv_wEjPSFzNN]

Let me know if you need any additional information.

Sorry, this link should work:

The files are inaccessible and says you need to provide permission to access them.

Does this link work? Sorry, don’t use Google Drive often.

https://drive.google.com/drive/folders/1A_8Ck6ya59ai0BNPpPYVYv_wEjPSFzNN?usp=sharing

Yes, that works.

Have you tried minimizing without the metal site? Immediate seg fault usually means infinite energy, so the first troubleshooting step I would take is to remove anything nonstandard to make sure the rest of the system is sensible, and then you know where the issue is.

Steepest Descents:
   Tolerance (Fmax)   =  1.00000e+03
   Number of steps    =         1000
Step=    0, Dmax= 1.0e-02 nm, Epot=  3.92826e+34 Fmax=         inf, atom= 9313

I get a seg fault, but it gives the reason why - you have something physically impossible near atom 9313. You’ve got a lot of unusual species in this system, many small molecules bound, and the box appears to barely fit the protein, so I’d suggest backing up and working first with only the protein in a suitable box of water, then slowly incorporating the nonstandard species (there are some sugars, an acetate, heme, etc) to see where the problem is.

Update: look at the geometry of the NAG residue. It’s completely twisted on itself. There is no hope of minimization resolving it. Zooming in on atom 9313 makes this apparent.

I suspected that any instability was coming from one of the nonstandard residues, but it was difficult to diagnose without the error message. Thank you for your help!