Average values of RMSD, Rg, SASA and h bonds: is it reasonable to calculate?

GROMACS version: 2023.3
GROMACS modification: No
Is it reasonable to calculate average values of such indictors as RMSD, Rg, SASA and number of H bonds over the entire length of the molecular dynamics simulation?
For example, can I write that “RMSD for protein was 0,35 ± 0,1 nm during 200 ns simulation” or conclude that average number of H bonds was 14,35 ±1,0?
RMSD according to the terminology is “root-mean-square deviation”, so I’m not sure that calculations of the average values are applicable for RMSD.

Yes, it is. These are dynamical quantities, for example the SASA of something will change as that something moves around. In general, I would say that quantities calculated on a single frame carry always little to no significance, as you are taking a picture of a single micro-state.

Regarding the RMSD, it depends on what you are calculating. For example, I would say that if you are interested into the RMSD of the backbone atoms of a protein, you want to take a look at how the plot behaves as a function of time. Is this stable/does it reach a plateau? If you think about it, reporting just the average and the RMSD won’t show if the system is i) oscillating around a plateau value with somehow large oscillations or ii) having somehow short oscillations but an average value that is drifting. At the end of the day, what do you want to show? I guess it really comes down to that.

So, if you are calculating ensemble average quantities, you have to be sure that your system is at equilibrium and then compute the quantity over time. Both the error associated and the length of the trajectory that you used are important to report, and may hint at how relevant something is.

What may be the best way to compare different trajectories based on the SASA, Rg, number pf H-bonds, etc? On the one hand, it may be difficult to draw conclusion only using graphs because they may be similar. On the other hand, calculation of the average values for the whole simulation may give less information about dynamics and fluctuations.