PBC effect on RMSD calculation

User discussions
, ,
1

GROMACS version:2024.1 (single precision)
GROMACS modification: No

image
image459×353 24.8 KB

So after my MD simulation like many other in the forum I have also encountered some pbc erros. But thanks to many users and developers I have been able to solve some of my problems. However I have tried many methods but couldnt fix a pbc effect (İn my opinion) after doing my RMSD calculation. In my MD videos I dont see any jump or difussion of the peptide however in my graph I see a sudden spike drop around 90 ns. I have used the fallowing codes:

  1. gmx_mpi trjcat -f step5_1.xtc step5_2.xtc step5_3.xtc step5_4.xtc step5_5.xtc step5_6.xtc step5_7.xtc step5_8.xtc step5_9.xtc step5_10.xtc -o MD_merged.xtc -settime

  2. gmx_mpi trjconv -s step5_10.tpr -f MD_merged.xtc -o MD_merged_center.xtc -center -pbc mol -ur compact

  3. gmx_mpi trjconv -s step5_10.tpr -f MD_merged_center.xtc -o MD_merged_center_noPBC.xtc -pbc nojump -ur compact -n all_index.ndx

  4. gmx_mpi rms -s step5_10.tpr -f MD_merged_center_noPBC.xtc -o rmsd.xvg -n all_index.ndx -tu ns

But it didnt seem to fix it. I would be greatfull for getting an explonation what I have been encountring with and how to fix it. Also is it normal for a graph to not start from 0 but around 1.5nm RMSD

2

With spike you mean the sudden drop to zero of the RMSD?

Generally, PBC problems go the other way around, you will see suddend spikes to very large values because the reference is broken across PBCs, commonly followed after a few frames by a drop again to pre-spike values when the image is again whole across PBCs.

Whether it makes sense or not that your rmsd starts from large values depends on the reference structure you are using and what part of the trajectory you are analysing. If you are using the first frame of the same trajectory then the RMSD must formally start from zero. The fact that you go back to zero after ~90ns might be a random fluctuation, but if the value is zero then I would say that this is nearly impossible and it’s most likely that you i) rebuilt the trajectory with trjcat incorrectly (e.g. concatenating trajs that are not consecutive) or ii) are using as reference frame for rmsd the frame at ~90ns

Edit: I think I spotted the problem. The gmx rms command uses as reference the frame contained in the -s reference file. Here you are using the tpr file you compiled to run the last trajectory, so correctly the first frame of the last part of the trajectory you glued together is distant zero from itself. Use as reference the frame you want as reference, not a random tpr, so if you want a comparison wrt the beginning of this simulation then use the firsr frame of step5_1.xtc (so its tpr should work).

3

thank you for your answer I think ı have doone it with this code:

1)#gmx_mpi trjconv -s step5_10.tpr -f MD_merged_center_noPBC.xtc -o start.pdb -dump 2)gmx_mpi rms -s start.pdb -f MD_merged_center_noPBC.xtc -o rmsd.xvg -n all_index.ndx -tu ns

and as you said (as ı understand) it started the graph from 0 which is the 0th frame and also the sudden spike drop is gone as seen below. the RMSD values are very high but ı gues the new (black graph) is biologicaly relevant one and correct one.

image
image1031×747 6.36 KB