I have completed a 5 ns MD run of my polymer (1 chain: 2162 atoms) in methanol (~36,000 CH3OH). I have checked the Pressure, Temperature and other properties during the MD running time and they seem to be stabilised.
I have checked the RMSD for the polymer chain (only) over the MD run w.r.t. the t0. The plot shows an increase at first (up to ~1 ns) and stability around ~0.3 nm for the rest of ~ 4 ns.
(A) My curiosity is about the RMSD calculation way. I have used,
(1) gmx trjconv : to convert PBC to no PBC and then get only polymer trajectory.
(2) gmx rms : to get RMSD of the polymer chain only.
As the polymer rotates and translates during the MD run, does it affects the RMSD values? If yes, how to remove these effects to lower the RMSD values?
(B) The polymer backbone has spiro connectivities and is almost rigid. It has side chains that fluctuate a lot and their movements are the primary concern of the study. I wonder how to get RMSD for the side chain and backbone separately?
(C) How to superimpose several geometries from the trajectory to generate an image?
If these need any other program and/or script to use, please let me know?
Any suggestions and comments are welcome.
Thank you for your time.
Quantities like temperature converge within a few hundred ps usually, so do not take these properties as any kind of indication that your sampling is sufficient. I dare say you need at least 100x more sampling for any polymer.
No, because this is what the least-squares fitting by gmx rms handles.
Create index groups for them and analyze them separately.
This is a question for your favorite visualization software. All of the common ones provide simple alignment tools.
Thank you for your reply. They are very helpful. I have a few queries,
Quantities like temperature converge within a few hundred ps usually, so do not take these properties as any kind of indication that your sampling is sufficient. I dare say you need at least 100x more sampling for any polymer.
I started the MD after equilibration, so they are supposed to be converged (unless very bad thing happens).
What are the properties or indications that I should look for to validate the MD results? How do I know that the total time frame is enough for MD?
“100x more sampling” does it mean 100*5ns total MD running time or more sampling with different initial geometries?
My “100x” comment was with respect to time. You need to consider the intrinsic time scales of motions relevant to your system. Running 100 5-ns simulations will show you nothing, as water has barely equilibrated (if it even has) and your polymer is still relaxing.
Convergence is assessed by time-independence of the properties of interest. These are going to be structural and dynamic properties of the polymer. Stuff like temperature and pressure will reach desired values quickly because they are forced to by the applied algorithms. How the system subsequently evolves under that statistical mechanical ensemble is what you are interested in.
Hello Justin,
Thank you for the comment.
I have already started a more extended time simulation after this 5 ns.
As far, I have understood that a 100 ns simulation is more meaningful than many small trajectories. secondly, it’s better to run a few long time trajectories.
Regards
Ranajit