Simulation Breaks the system (separates both the chains of the Antibody)

GROMACS version:
GROMACS modification: No

Hi @jalemkul
I have run the MD simulation (protein-lignad) by following your protocol ( Also at the end of simulation, I recentered and rewrapped the coordinates using gmx trjconv command. In MD simulation movie it gives weird outcome. I ran MD simulation for 200 ns.

Energy minimization graph is as follow.

since this portal doesn’t allow to upload video, please follow the link to see the video (protein_ligand_movie.mp4 - Google Drive). Also, I am attaching few frames of it.

At the begining:

after some time (around 50 ns):
At near 100ns
at the end of simulation

Throughout the simulation, both of the chains of Antibody keep joining and separating. Why is this happening?


This is a simple PBC effect. Use gmx trjconv to center on one chain (not the default “Protein” group, you need to create an index group corresponding to one of the protein chains).

Thank you @jalemkul

First, Do I need to modify the index file by adding a group having atoms of Heavy or Light chain only. Then run the production MD commands.

gmx grompp -f md.mdp -c npt.gro -t npt.cpt -p -n index.ndx -o md_0_10.tpr

gmx mdrun -deffnm md_0_10

after MD production, gmx trjconv and then select one chain to be centered.

If not so could you please help me how to do that?


After MD simulation, I created a group (light_chain) in index.ndx file and run the following command.
gmx trjconv -s MD.tpr -f MD.xtc -n index.ndx -o MD_center.xtc -center -pbc mol -ur compact

I selected the following options

select a group for centering-> light_chain
select a group for output-> system

Upon visualization same result. it separates both the chains apart.

Hi Users, @jalemkul
Still struggling with this problem. Any help will be appreciated.


Please render the box itself in your image.

Did you mean MD movie having solvent, ion and box?
In Chimera, I could’t find a way to get box boundaries. So, I tried it by uploading MD.gro and MD_recenter.xtc without deleting water/SOL molecules. Please find the url.

Your simulation box is too small to properly accommodate your protein complex, so one chain hits a periodic boundary and jumps. This is not resolvable via trjconv and you need to check your simulation carefully for minimum image violations.

Thanks for the suggestion.

How to determine the appropriate size/dimension of the box? I have used the box of same dimensions/size for a protein-protein (macromolecule) MD simulation and that worked perfectly. However, in Protein-Ligand MD simulation I’m facing this issue.

I am worried if I choose bigger box than required, lots of water molecule and ions might make simulation unnecessary heavy and this might impact the results.

You need a suitable spatial buffer around the solutes to avoid minimum image violations anyway, so if the solute is large, so too will your system. You can use something more efficient than a cubic box, like a rhombic dodecahedron, but editconf -d 1.0 is typical to get an appropriate box size.

I also have a similar problem. My system consist of two protein chains (protein-protein complex). After doing the md I see on trajectory that chains are divided and move out of box. I used procedure as for „protein in water”. I ask for help: on which stages and what changes I should make in order to exclude such situation: during protein-protein complex preparation, just before md or after md?