Rmsd drops at the last frames

Hi all,

I calculated the RMSD for my system which contains 4 monomers. It starts from around 1nm and goes down at the last 10 ns of the 50 ns simulation to near zero at the last frame. Does it mean the system is not well equilibrated and I should extend the simulation or there might be something wrong with the command I used or the PBC effect is not removed properly?

The RMSD command is:
gmx rms -s new4-c.gro -f new4-c.trr -tu ns -o rmsd.xvg -n index.ndx (backbone,backbone)

And the commands I used for removing PBC are:
gmx trjconv -f md.trr -s md.tpr -o new1.pdb -b 0 -dump 1

gmx trjconv -f new1.pdb -s md.tpr -o new1-w.pdb -pbc whole

gmx trjconv -f new1-w.pdb -s md.tpr -o new2-cl.pdb -pbc cluster -n index.ndx

gmx trjconv -f md.trr -s md.tpr -o new1.trr -pbc whole

gmx trjconv -f new1.trr -s md.tpr -o new2-cl.trr -pbc cluster -n index.ndx

gmx trjconv -f new2-cl.trr -s new2-cl.pdb -o new3-j.trr -pbc nojump

gmx trjconv -f new3-j.trr -s md.tpr -o new4-c.trr -pbc mol -center yes -n index.ndx

(And the same 4 commands to process the .gro file)

Any help would be appreciated.
Thank you.

Hi Marko,

This sounds suspicious - I would rather assume that one monomer eneded up in a strange position to begin with and then jumped some periodic boundary in the end of the simulation.

If I understand correctly, you use the first frame of your simulation as a reference for the RMSD calculation. If all the pbc correction goes well, this should be exactly the same as the simulation frame and have RMSD = 0.

Visualising the trajectory will help, vmd and pymol should both be able to load in trajectories, if not trr they can read your xtc.

There shall be no jumps in MD.

Thank you for your response.
I visualized the 10 last ns of the trajectory and it looked fine. It had enough movements not to result in a drop in RMSD and RMSD~0 at the last frame. Also, I have not noticed a monomer being separated or having a strange position (As far as I controlled).
As you mentioned the first frame of the production run was the reference for RMSD. So, according to what you explained, since the RMSD is ~0 it shows that the pbc correction was done well? Then whenever we have the pbc removed properly we should get RMSD=0? Does it mean I should calculate RMSD with the unremoved pbc? In other words, suppose the pbc was corrected properly. How can I calculate the RMSD not to give me a drop and ~0 at the end?
Thank you.

RMSD = 0 at t = 0 is naturally going to happen if the reference for the calculation is the starting structure. What is more important is what happens at the next snapshot. You immediately have a large jump in value. Whatever is causing this should be apparent upon viewing the simulation by loading the reference coordinate file and then the trajectory being analyzed in e.g. VMD. Something dramatic should happen, and you need only look at the first two frames to focus in on what the issue might be.