hi ,

20ns is a really less time frame for MD simulations. Hence your conformations did not take a specific shape or are in a correct orientation. You can validate this from the RMSD plot you get after running MD. If fluctuations are present in the plot at the value of 20ns then clearly the structure needs more time to get stable. Try running MD for a longer time , generally 100ns or 50ns at least. Observe where you get the stability in all your plots.

However, if choosing 20ns was very specific to your study then there can be a different explanation but generally no structure is stable at 20ns. Hope this helps :)

You can reach out to me if the problem persists.

I have faces with similar glitches when I didn't center and fit my system. Have you processed your system?

Looks like your protein drifted into the periodic boundary box, making part of it to jump to the other side of the box.

You need to specify NoJump before exporting the coordinates into the viewer program. Consult the manual of the specific MD analysis program you use for the exact synthax to do this. e.g. https://docs.mdanalysis.org/stable/documentation_pages/transformations/nojump.html

You need to remove remove the pbc effect. To do so, you can use this command :

gmx trjconv -s md_file.tpr -f md_file.xtc -o md_file_noPBC.xtc -pbc mol -center