Can you help check RMSDs trends of 100 ns production mdrun?

[minghao2121]

GROMACS version: 2021.1
GROMACS modification: No
Hello,
I did 100 ns production mdrun followed by the wonderful tutorial (Lysozyme in Water) using my own protein. I made a figure for wide-type (wt_rmsd, backbone RMSD by Ref: equilibrated) and one-mutation (mi_rmsd, backbone RMSD by Ref: equilibrated) and two-mutation (miv_rmsd, backbone RMSD by Ref: equilibrated) together as attached here.

I, as a beginner of GROMAC MD, have some questions about this figure.
(1) For these 3 series, is 100 ns enough? It seems mi_rmsd has an increasing trend and miv_rmsd has a jump at about 92 ns.
(2) If 100 ns is not enough, should I continue another 200-ns MD?
(3) Generally, what is the criteria (or thumb of rule) for a good MD simulation indicating the simulation time is enough?
(4) What can I learn from this figure? wt_rmsd is more stable than mi_rmsd and miv_rmsd?
(5) Can I do some following structure analysis based on the 3 series rmsd? I mean are these 3 series RMSD indicating my MD is stable and can do the following structure analysis?
(6) Generally, how long (300 ns is enough?) should I perform production mdrun for “enzyme in water” in order to figure out the conformation changes caused by the single-point or double-point mutations by comparison with wide-type?

Thanks for your help!
Ming

[jalemkul]

@minghao2121

(1) For these 3 series, is 100 ns enough? It seems mi_rmsd has an increasing trend and miv_rmsd has a jump at about 92 ns.
(2) If 100 ns is not enough, should I continue another 200-ns MD?

You have to simulate long enough to observe converged sampling of the relevant dynamics. We can’t possibly tell you whether the simulation is long enough. If you’re interested in fast side chain dynamics, yes, probably. If you’re interested in domain motions or folding, no.

@minghao2121

(3) Generally, what is the criteria (or thumb of rule) for a good MD simulation indicating the simulation time is enough?

When all relevant quantities are statistically invariant over time.

@minghao2121

(4) What can I learn from this figure? wt_rmsd is more stable than mi_rmsd and miv_rmsd?

Stability is a thermodynamic concept. RMSD can be a proxy for this, but don’t read too much into it. You’re making the assumption that your mutants will start from the same tertiary structure in spite of the mutation; that may not be true. Your result means the WT deviates less than the mutants from the initial structure, but whether or not that reflects stability is not 100% clear. RMSD is a fairly crude quantity.

@minghao2121

(5) Can I do some following structure analysis based on the 3 series rmsd? I mean are these 3 series RMSD indicating my MD is stable and can do the following structure analysis?

Sure, and the analysis itself will tell you if the simulation is adequately converged. Even with a flat RMSD, other quantities can be varying.

@minghao2121

(6) Generally, how long (300 ns is enough?) should I perform production mdrun for “enzyme in water” in order to figure out the conformation changes caused by the single-point or double-point mutations by comparison with wide-type?

Depends entirely on how large those conformational changes are. There is honestly no way to answer this question in any general way.

[minghao2121]

Thanks Justin for your details answers. My many questions are solved now. I will try to do longer simulation and also do the structural analysis to see if I can get the interesting conclusions.

Best,
Ming