Getting only 3 columns in xvg file error while using gmx bar

User discussions
1

GROMACS version: 2024.1
GROMACS modification: No

Hi everyone, I ran gromacs simulation with free-energy option as yes. The sequence was: minimization >> equilibration (NVT) >> production (NVT).
After completing the production run, I collect all the dhdl.xvg (named as feprun$(lambda_value).xvg) files for all lambda values in a folder and run gmx bar command.

gmx bar -f *.xvg -o -oi -oh

This command throws the following list of errors (only a few lines pasted here):

Only 3 columns on line 90247 in file feprun0.xvg
Only 3 columns on line 149189 in file feprun0.xvg
Only 1 columns on line 149897 in file feprun0.xvg
Only 1 columns on line 176646 in file feprun0.xvg
Only 3 columns on line 181269 in file feprun0.xvg
Only 1 columns on line 224137 in file feprun0.xvg
Only 3 columns on line 251481 in file feprun0.xvg
Only 3 columns on line 336611 in file feprun0.xvg
Only 1 columns on line 368063 in file feprun0.xvg
Only 1 columns on line 371324 in file feprun0.xvg
Only 3 columns on line 414060 in file feprun0.xvg
Only 1 columns on line 426038 in file feprun0.xvg
Only 1 columns on line 481909 in file feprun0.xvg
Only 1 columns on line 518428 in file feprun0.xvg
Only 1 columns on line 521959 in file feprun0.xvg
.
.
.
Segmentation fault (core dumped)

I opened the file but there are four columns on the respective line numbers. The production run for each lambda value is particularly long = 180ns. Due to this, the first column occupies more number of spaces, but I do not think that should matter as gromacs is used to run simulations that are typically as long if not more.

I request you to help me in figuring out this error.

Thanks,
Chaitanya

2

Could you post the xvg file somewhere and make a link to the file? Then we could have a look at it and see if we find anything strange.

3

Hello Magnus,

Thanks for your reply.
Please find the files in the link below:

xvgfiles

Please let me know if the link isn’t working or if there are permission issues.

Chaitanya

4

The files look OK to and I can run gmx bar on them without any errors. I used the current GROMACS development branch, but there shouldn’t be any differences.

So, there might be some problem with your installation. Are you sure you are running 2024.1 for the analysis? Are you running the simulations and analysis on the same computer? That shouldn’t be necessary, but perhaps there is some filesystem issue. Otherwise, I don’t know what could be the problem.

Anyhow, in the future I’d recommend running without separate-dhdl-file, especially when you are writing so often. If you’d need ascii output you can extract that later

5

Hello Magnus,

Thank you responding again.
I am sure I am using gromacs 2024.1.

But, I think you pointing out that there must some problem with the installation is correct. I say this for two reasons:

  1. I tried to use gmx bar command on the same files but a different machine. And it worked just like it did for you, without any issues.
  2. There was a issue with the machine due to which the simulations were interrupted. We had to reinstall the Nvidia drivers and CUDA toolkit. After that I had not reinstalled the gromacs though, since the drivers and the toolkit reinstalled were exactly the same version as earlier.

However, I will reinstall gromacs to be on the safe side. Also, will you recommend rerunning the simulation to be sure the results are correct?

Also, can you please elaborate on why I should run without separate-dhdl-file option?

Thanks again for your help.
Chaitanya

6

Those results seemed fine, so unless you see any particular errors in any of the other files, I don’t see any reason to rerun the simulations.

separate-dhdl-file writes an ascii (human readable xvg file) with your free energy output. That file is a lot larger than the binary ener.edr file, where the same data is written if you don’t use separate-dhdl-file. You can re-create a dhdl file from the energy file, but you can also you use the energy file as input to, e.g., gmx bar.

7

The results look fine to me. Just wanted an expert opinion on this.

Thanks for explaining the separate-dhdl-file option.

Regards,
Chaitanya

8

Hi Magnus,

I ran into an error while re-installing gromacs.

After I complete running the cmake command and do make, I am getting an internal compiler error, while trying to install with or without gpu.

Can you help me where to look in this regard?

Did not get this error previously.

Regards,
Chaitanya

9

I’d probably need more information than that. I’d suggest starting another thread to make the problem easier to discover by its topic. Hopefully someone can help you, at least.

10

Thank you Magnus.

Chaitanya