GROMACS version: 2023.1
GROMACS modification: No
Hello,
I performed umbrella sampling simulation and when i typed this command:
gmx wham -it tpr-files.dat -if pullf-files.dat -o -hist -unit kCal
i get a profile.xvg file
profile.xvg (3.9 KB)
which has not values at the second column, but instead it reports nan as you can see at the file attached.
md_umbrella.mdp (2.0 KB)
npt_umbrella.mdp (2.3 KB)
I also upload the mdp files i used for equilibration and production.
Has anybody else ever faced this problem?
I also get this message in terminal when i type the command.
Wrote histo.xvg
Getting initial potential by integration.
Warning, poor sampling bin 197 (z=4.2514). Check your histograms!
Warning, poor sampling bin 198 (z=4.26911). Check your histograms!
Warning, poor sampling bin 199 (z=4.28682). Check your histograms!
Initialized rapid wham stuff (contrib tolerance 1.63934e-08)
Evaluating only 0 of 12200 expressions.
Switched to exact iteration in iteration 1
1) Maximum change -1.000000e+20
Converged in 2 iterations. Final maximum change -1e+20
It’s difficult to give specific advice without more details. From the warnings it seems like there is insufficient sampling. How many umbrella windows do you have? What does histo.xvg look like?
Just looking at the mdp file there are no apparent reasons why the umbrella sampling would be insufficient. But there are a few other warning bells that ring:
- Why are you using five temperature coupling groups? I would argue that you should use as few as possible. I don’t see why you would need more than one.
- Parrinello-Rahman with
tau_p = 1.0 might cause system instabilities. I’d recommend using a c-rescale barostat. I would also recommend considering compressibility = 4.5e-5 0 when you are using umbrella sampling along the Z dimension.
Dear Prof.,
I also attach the histo.xvg file.
histo.xvg (158.2 KB)
I dont think that happens due to the five temperature coupling groups or the instability of the system, cause other umbrella simulations with the same options provide accurate results.
I have 60 umbrella windows. After i got 500 .gro files with the pulling biasing force, i analyzed umbrella100.gro-umbrella400.gro with increment of 5.
Thank you for the response!
I agree. I don’t think the problems with the analysis are due to the thermostat and barostat changes. Those were just unrelated recommendations.
The histograms you attached look fine. So, I must admit I can’t say what is going wrong. I presume the data set is very large, so it’s difficult to share it, right?
The data set is way too large in terms of memory, but the results are fine. When i open the .xtc and .gro files, everything runs normal. I will study it further and come up with a solution.
Thank you for your time!