High Fluctuations in DNA RMSD Analysis After PBC Correction

User discussions
1

GROMACS version: 2024
GROMACS modification: No

Hi fellow GROMACSers,
I am running a molecular dynamics simulation of DNA and analyzing the RMSD to evaluate its stability. However, I observe significant fluctuations in the RMSD plot, even after applying periodic boundary condition (PBC) corrections. The RMSD values range from 0 to 6 nm without showing clear convergence.

Here is the workflow I followed:

I corrected for PBC using the following command:

gmx trjconv -s md_0_1.tpr -f md_0_1.xtc -o output.xtc -pbc mol -ur compact -center

I selected the “DNA” group for centering and the “System” group for output.

I calculated the RMSD using the command:

gmx rms -s md_0_1.tpr -f output.xtc -o rmsd.xvg -tu ns

For both the least-squares fit and RMSD calculation, I selected the “DNA” group.

I visualized the RMSD plot using xmgrace rmsd.xvg.

Despite following this process, the resulting plot shows large fluctuations.

I would appreciate expert advice on potential causes for these fluctuations and the best way to address them. Thank you!

RMSD
RMSD913Ă—625 4.25 KB

2

Try

gmx trjconv -s md_0_1.tpr -f md_0_1.xtc -o md_whole.xtc -pbc whole
gmx trjconv -s md_0_1.tpr -f md_whole.xtc -pbc nojump -o md_nojump.xtc

and possibly a gmx trjconv -s md_0_1.tpr -f md_nojump.xtc -c -o md_center.xtc as well.

And read Terminology - GROMACS 2024.5 documentation. You might want different PBC removal procedures for a trajectory you want to analyze compared to one you want to visualize.

1 Like
3

Hi MagnusL,

Thank you for your helpful advice! I followed your recommended workflow to preprocess my DNA trajectory:

gmx trjconv -s md_0_1.tpr -f md_0_1.xtc -o md_whole.xtc -pbc whole  
gmx trjconv -s md_0_1.tpr -f md_whole.xtc -pbc nojump -o md_nojump.xtc  
gmx trjconv -s md_0_1.tpr -f md_nojump.xtc -center -o md_center.xtc

The corrected trajectory looks clean in VMD: the DNA stays centered, no PBC jumps, and both strands remain paired. I’d like to ask two follow-up questions about analyzing and stating a new simulation:

  1. Do the RMSD/RMSF plots below suggest stable DNA dynamics? The RMSD fluctuates between 0.2–0.6 nm, and RMSF is highest at the termini (attached images).

  2. I want to use the output of this simulation to start a new MDS.
    Would you recommend using the last frame directly, or is it better to select a cluster centroid or the structure closest to the average?

Your insights would mean a lot—thanks for making this forum such a great resource!

new.rmsd
new.rmsd980Ă—616 5.65 KB

rmsf
rmsf980Ă—616 4.13 KB

4

I know too little about DNA dynamics to say whether it’s stable or not. But if you say that the strands remain paired (when visualized) then it’s probably OK. It’s diffcult to say whether the RMSD will continue increasing.

Would you like to start a new simulation, to collect more statistics, or would you like to extend this one? If you believe that the system is in equilibrium, it should not matter in principle. But since the RMSD is higher at the end of the simulation you have already done, you need to ask yourself what kind of additional data you want. Better statistics? Or see the behaviour over a longer period of time? Perhaps it’s a good idea to start one or two more from the initial conditions, and extend this one.

5

I’m working on simulating the interaction between a peptide and DNA, following a protocol from the literature. The authors describe a two-step approach:

  1. Pre-simulation: Running separate simulations for DNA (100 ns) and the polycation (50 ns) in water.
  2. Complexation simulation: Using the final structures from these pre-simulations as starting conformations for the main polycation-DNA complex simulation.

I have run the separate pre-simulations in water. My question is: How should I properly use the final structures from these pre-simulations as input for the complexation simulation?

6

Then I would suggest starting from the end (confout.gro) of the simulations you have run.

Edit: I’m not sure exactly how you would be combine these two simulations. Don’t they say anything about what they have done? If you want to preserve the state from the equilibrations, you would preferably like to retain one or two layers of hydration around the DNA and the polycation (peptide?), combine them at a suitable distance in a simulation box and re-solvate the system. Why don’t you contact the authors and ask how they did?

7

Thank you for your suggestion! The authors did not provide any specific details regarding the method they used to combine the two simulations. I would have preferred to follow their exact protocol, but it seems they left those details out.

Thank you again for your help and advice!