I am trying to run a 34 amino acid peptide through very high temperatures to denature it. Other publications have done this at 600 K so I am trying to do the same. To do so I simply edit the following lines in the {md, nvt, npt}.mdp files:
; Temperature coupling is on
tcoupl = V-rescale ; modified Berendsen thermostat
tc-grps = Protein Non-Protein ; two coupling groups - more accurate
tau_t = 0.1 0.1 ; time constant, in ps
ref_t = 600 600 ; reference temperature, one for each group, in K
After running the simulation for two days on 48 nodes of a computational cluster, I only get about 80 picoseconds when I wanted 200ns. And after looking at the radius of gyration of the 80 picoseconds I get, the peptide does not seem to be denaturing. I was wondering if what I am doing is correct or if there is another way to better modify the temperature? Also why does the simulation take so much longer than at normal temperatures?
Thanks! Please let me know if I should provide more information.
Not much will happen in 80 ps. The better question is why you got 80 ps when you were expecting 200 ns. Did the run crash? The temperature itself will absolutely not change performance. If you're getting serious performance degradation, something else is going on.
Also keep in mind that you're subjecting the system to extreme conditions, for which most force fields were not designed. You will have to use a pretty small time step for this to be stable, and NPT may never be stable; most high-temperature MD is done in NVT.
Thank you Justin. The run didn't crash. It terminated because it exceeded the -maxh of 48 hours that I set which would normally be enough for a 200ns simulation.
Okay so I will try it again doing only NVT equillibration.
Would you have an idea of what an appropriate time step could look like?
Right now i have:
dt = 0.002 ; 2 fs
Thanks!
1 fs or lower. If you get 200 ns in 2 days, you must have really good hardware :)
Hello, I have completed a bit over 100ns of the simulation using your suggestions. So the problem of reduced performance is gone, however the radius of gyration appears to be fairly stable at a size very similar to the one it has at a normal temperature. At this point I am unsure what the problem could be. Please find attached my nvt.mdp and md.mdp files, as well as plot of the radius of gyration. Would you have any idea what could be missing? Is there a way to plot the temperature of the system over time to check that it's right?
Thanks!
The Rg is rather bizarre; it's becoming more compact. How big is your box? Is there adequate room for a protein of this size to unfold?
All energy terms, including temperature, are written to the .edr file.
I made the box using the editconf tool with the -bt dodecahedron and -d 1.0 option.. The peptide is a fairly unstructured 34 aa chain.
Strange. Couple of questions: what parameters did you use when performing energy minimization and equilibration? Was system stable before you started production run MD? Are you using periodic boundary conditions?
Perhaps try to plot temperature and pressure over time and check if it is stable. You can plot many parameters over time (including pressure and temperature) using tool gmx energx. You can also try to visualize simulation trajectory with an application such as VMD:
http://www.ks.uiuc.edu/Research/vmd/
You might see what is wrong. If your peptide is fairly unstructured, you might need to increase box size, but at the moment, I don't think this is the main problem.
- Badges
- Science topic