Recentering long, tetrameric enzyme simulations properly

GROMACS version: 2020.4
GROMACS modification: No

Hi all,

I’ve ran a 60 ns simulation of a tetrameric enzyme (331 residues per monomer) with its cofactor present (LDHA:NADH, roughly 20k atoms, 1.5 nm distance from box edge) and I’m having trouble recentering the coordinates properly. Is there a default way to handle such systems?

What I’ve tried so far is to use the -center -pbc mol -ur compact flags which splits the tetrameric enzyme into two dimeric systems at the edge of the unit cell (some residues outside), using the flag -pbc res has similar results and -pbc atom gives a result that looks like the initial uncentered structure.

I spoke with the author of a relevant MD paper (same tetrameric enzyme-ligand system size, same simulation length) to see how they tackled the issue and they said they remember playing with different flags to solve the same issue so I’ll read round further but was curious if anyone has experience with these kinds of systems.

Thanks in advance
Sam

Create an index group corresponding to one protein chain and choose it as the group to center using gmx trjconv.

Thank you, this has worked.

One minor question, will this give me superfluously high RMSD values? I tried to calculate RMSD values for backbone atoms and the values start around 6.4 nm as opposed to 0.

Thank you for helping
Sam

(Edit: I think I’ve solved this, I used the starting centered gro file (I created with gmx trjconv -dump 0) instead of a .tpr file)

Indeed, the reference state needs to be similarly centered so that a meaningful fitting can be done without any chance of jumps across PBC.