GROMACS version: 2023.3-foss-2023a-CUDA-12.1.1-PLUMED-2.9.0
GROMACS modification: No
Hi everyone,
I’m working on a simulation involving a protein dimer embedded on a lipid bilayer. After running the MD simulation, I attempted to remove periodic boundary conditions (PBC) and center the system on the protein, but I’ve run into a persistent issue: The protein dimer keeps flipping or jumping across the simulation box (attached pics), even after trying several gmx trjconv strategies.
-pbc whole -center
-pbc mol -center
-pbc mol -center -nojump
Tried different centering groups (protein vs system)
None of these managed to produce a trajectory where the protein stays centered and the bilayer remains intact — it still jumps or flips across frames.
I’d greatly appreciate any advice, best practices, or working workflows for this kind of protein-membrane system visualization. I’ve searched around but haven’t found a method that solves this cleanly.
Thanks in advance for your time and help!
Best,
Colin
I am not sure what you have tried so far. However, I would try the following:
Create an index file (e.g., center.ndx) with a selection group that contains only one atom at the binding interface . This could be for example an atom of a central side chain interacting with the membrane. Additionally, create another selection group that contains only the protein (or protein+membrane).
Call gmx trjconv -f .xtc -s .tpr -pbc atom -n center.ndx -center -o atom_center.xtc. Select the atom for centering, and your system as output.
Call trjconv again: gmx trjconv -f atom_center.xtc -s .tpr -pbc whole -n center.ndx -center -o whole_center.xtc Now you select the group only with the protein (or protein+membrane) for centering and the group you like for the output.
This workflow will (hopefully) center your region of interest in the simulation box and eliminate sudden PBC jumps. If the first attempt fails, try selecting a different atom at the binding interface for the initial centering step. To center the membrane, choose an atom within the membrane core—such as the terminal carbon of a lipid chain—for that first centering.
I can’t guarantee it will work for every system, but in my experience it produced reliable results for systems with relatively short box vectors.