Dear Archchana Ganeshalingam

The RMSD plots attached here are not stable at all. RMSD vs Time trajectories will be stable if you notice a stable plot between these two. Try to run your simulations for longer period of time and subsequently check for the convergence by plotting RMSD vs Time.

Check my publication below where you can get an idea of stable adn unstable RMSD plots.

https://doi.org/10.1002/prot.26022

Hope it will work for you.

Whether the plots are stable or not depends on an arbitrary definition of stability. Besides, flat RMSD plot on its own means nothing: past RMSD is not a predictor of future RMSD.

i would rather look at the snapshots with the highest values of RMSD: which parts of the simulated system contribute most and why? Large fluctuations in RMSD can be due to rigid-body motions, such as, in the case of proteins, domain-domain motions around rigid hinge, and these motions may be important for the function of the protein.

Anyway, i would divide the trajectory into two halves, calculate average RMSD for each, and if the two averages are within two standard deviations, i would consider the two halves similar in terms of RMSD.

Longer MD simulation does not imply better 'convergence.'

Martin Klvana

Read my answer carefully, what I have written earlier. Carry out multiple MD simulations and subsequently check for convergence, which means after few ns, plot RMSD vs Time plot, will give a idea about stability. RMSD itself does not contribute for stability, there are several other parameters responsible for stability like Rg, SASA, FES, structural snapshots.

Yes, it is true that past RMSD is not a predictor of future RMSD, but flat RMSD means nothing, it is not true.

A stable RMSD is kind a first indicator of stability. Does that mean it's stable? Not necessarily. But if you analyze native contacts, hydrogen bonding, time evolution of secondary structure , salt bridges networks etc, . you can put together an statement about your MD simulations that govern secondary and tertiary interactions. This is the real value in a MD simulation.

You must perform multiple 100 ns or more simulations. A single trajectory is not necessarily representatition of real change in the system and may not even be converged.

Then and only then you can get an evidence to convince the peer reviewers.

Hello, I have run MD of my complex in triplicates and the RMSD deviation is very high. How can I interpret these kind of results.