Restarting the simulation

User discussions
1

GROMACS version:2022.5
GROMACS modification: No
Hi,

I was running a 100 ns membrane+protein+ligand simulation job in HPC using SLURM. My job ran out of time and got killed approximately around 92 ns.

I tried to restart the simulation so as to complete the remaining 8 ns simulation using the following command:

gmx_mpi mdrun -s md.tpr -cpi md_prev.cpt -v -append -deffnm md

Then the simulation got over successfully. However, when I checked md.log file i can see the follows:

DD step 46263999 load imb.: force 79.1% pme mesh/force 1.054
Step Time
46264000 92528.00000

Energies (kJ/mol)
Bond U-B Proper Dih. Improper Dih. CMAP Dih.
5.34296e+04 2.43940e+05 1.84763e+05 4.48997e+03 -1.78181e+03
LJ-14 Coulomb-14 LJ (SR) Coulomb (SR) Coul. recip.
3.82478e+04 -2.64975e+04 3.25754e+05 -5.84548e+06 2.03223e+04
Potential Kinetic En. Total Energy Conserved En. Temperature
-5.00282e+06 1.12946e+06 -3.87336e+06 7.79489e+06 3.03070e+02
Pressure (bar) Constr. rmsd
-8.74943e+01 4.64061e-06

Writing checkpoint, step 46264500 at Wed May 29 17:44:03 2024

4 2.46595e+05 1.85823e+05 4.56233e+03 -1.95436e+03
LJ-14 Coulomb-14 LJ (SR) Coulomb (SR) Coul. recip.
3.87148e+04 -2.75305e+04 3.24986e+05 -5.84069e+06 1.62945e+04
Potential Kinetic En. Total Energy Conserved En. Temperature
-5.00036e+06 1.13140e+06 -3.86896e+06 7.79472e+06 3.03591e+02
Pressure (bar) Constr. rmsd
1.82992e+01 4.65514e-06

DD step 46278999 load imb.: force 4.4% pme mesh/force 1.029
Step Time
46279000 92558.00000

Energies (kJ/mol)
Bond U-B Proper Dih. Improper Dih. CMAP Dih.
5.35400e+04 2.45639e+05 1.85842e+05 4.61183e+03 -1.92825e+03
LJ-14 Coulomb-14 LJ (SR) Coulomb (SR) Coul. recip.
3.85054e+04 -2.72041e+04 3.24764e+05 -5.84297e+06 1.61553e+04
Potential Kinetic En. Total Energy Conserved En. Temperature
-5.00305e+06 1.13006e+06 -3.87299e+06 7.79500e+06 3.03230e+02
Pressure (bar) Constr. rmsd
2.64827e+01 4.68069e-06

From the above I can see that from step 46264000 and time 92528.00000 the trajectory jumped to step 46279000 and time 92558.00000. Later it keeps writing the frames till

DD step 46332999 load imb.: force 3.0% pme mesh/force 1.026
Step Time
46333000 92666.00000

Energies (kJ/mol)
Bond U-B Proper Dih. Improper Dih. CMAP Dih.
5.37965e+04 2.45166e+05 1.86201e+05 4.61486e+03 -2.18204e+03
LJ-14 Coulomb-14 LJ (SR) Coulomb (SR) Coul. recip.
3.82651e+04 -2.70649e+04 3.25583e+05 -5.84266e+06 1.62281e+04
Potential Kinetic En. Total Energy Conserved En. Temperature
-5.00205e+06 1.13090e+06 -3.87115e+06 7.80841e+06 3.03456e+02
Pressure (bar) Constr. rmsd
-9.84917e+00 4.64059e-06

And then it automatically jumps back to the following step and time

DD step 46264999 load imb.: force 92.0% pme mesh/force 1.020
Step Time
46265000 92530.00000

Energies (kJ/mol)
Bond U-B Proper Dih. Improper Dih. CMAP Dih.
5.31246e+04 2.45763e+05 1.85537e+05 4.52282e+03 -1.92235e+03
LJ-14 Coulomb-14 LJ (SR) Coulomb (SR) Coul. recip.
3.83409e+04 -2.73014e+04 3.27261e+05 -5.85147e+06 2.04276e+04
Potential Kinetic En. Total Energy Conserved En. Temperature
-5.00572e+06 1.12950e+06 -3.87622e+06 7.79479e+06 3.03080e+02
Pressure (bar) Constr. rmsd
7.18260e+01 4.63354e-06

To clear my doubt I ran gmx check on md.trr file and got the following output:

Command line:
gmx check -f md.trr

Checking file md.trr
trr version: GMX_trn_file (single precision)
Reading frame 0 time 0.000

Atoms 426608

Last frame 1000 time 100000.000

Item #frames Timestep (ps)
Step 1001 100
Time 1001 100
Lambda 1001 100
Coords 1001 100
Velocities 1001 100
Forces 1001 100
Box 1001 100

I think that I might have duplicated frames in my trajectory but after looking at the output from gmx check I am a bit confused. I want to know how can I get rid of the duplicated frames and why such a thing happened?

Also after the simulation ended I am not able to find md.xtc file. It appears that this was never created. How can I move ahead with analysis if I do not have an md.xtc file. Can some one help me with specific command to get the xtc file.

Many thanks in advance

2

I need some help. I will be grateful if any expert can clear my doubt.

Thanks and regards

3

If you didn’t set nstxout or nstxout-compressed in the .mdp file, the trajectory file would not be written, and it is hard to make further analysis.

4

Hi,

I solved the problem of generating md.xtc file by using the following command:

gmx trjconv -f md.trr -s md.tpr -o md.xtc

I have the md.xtc file now but I am still not clear about the discrepancy in the frames. Can someone guide me with that please.

Thanks

5

From gmx check it doesn’t seem like there are any duplicate frames.

6

Many thanks for replying. I have a small follow-up question though. I was just wondering why the steps and time in my md.log jumps ahead and jumps back? Could you please answer this. Many thanks for your help.

7

I don’t know. It almost looks like there are two simulations in parallel writing to the same log file, but there should be checks against that - if the file system is working correctly.

It’s very difficult (for me) to diagnose your problem afterwards. As long as the trajectory files and energy files are OK it shouldn’t be a problem. If it keeps happening it would be interesting to try to understand what the problem is.

8

Many thanks for sharing your views on this.

best regards