The beads detached from the molecule in coarse-grain MD simulation

Dear all,

I am trying to perform a coarse-grain MD simulation of double strand DNA. The coarse-grain model is 3SPN2. I used implict solvent and ions and periodic boundary conditions under NVT ensemble. After the simulation was finished, I loaded the trajectory into VMD. Strangely, the DNA looked ok in the first hundred frames but thereafter some beads moved far from DNA chains, looking like being detached. However sometimes it was intact, just as follows:


The box is:
-38.870000 38.870000 xlo xhi
-38.870000 38.870000 ylo yhi
-38.870000 38.870000 zlo zhi

I found that some coordinates of trajectory were larger than 38.87. In the docs of dump command, it notes that the coordinates of an atom written to a dump file may be slightly outside the simulation box. But I feel it’s abnormal in my case, because I think the beads should not have moved away from the DNA chain.

I’m very confused about this. Could you give me some guidance? I attached input and configuration file. Thank you very much in advance. (2.6 KB) (7.2 KB)

What you see is a side effect of outputting “wrapped” coordinates (“x”, “y”, “z”). Your entire object is moving through the box and due to periodic boundaries atoms leave the box and their periodic images enter. The visualization would look different, if you would output “unwrapped” coordinates (“xu”, “yu”, “zu”) instead or also output the image flags (“ix”, “iy”, “iz”) in addition to the wrapped coordinates. In the latter case, VMD will reconstruct the unwrapped coordinates when reading the trajectory.

Thank you very much. I used “dump mydump all custom 5000 traj.lammpstrj id type xu yu zu”. It works. Many thanks!