GROMACS version: 2024.4
GROMACS modification: No
Hi all,
Just posting to this forum because after running a triplicate of coarse-grain lipid bilayer systems with added proteins, I was surprised with very uncharacteristic behaviour from one of my replicates. Two out of my three systems in the triplicate ran as expected, in that, the box size fluctuated but roughly remained quite close to the input box dimensions, my proteins moved around freely in the box and stuck to the membrane as expected, and the membrane observed the curvature I’ve grown to anticipate from such membrane x protein systems.
Nonetheless, I have a “problem” replicate here, where despite the simulation equilibration steps running correctly (no errors encountered), the system is exhibiting very uncharacteristic behavior:
- The box dimensions are drastically different than my input box’s parameters; it shrunk by 7 nm in the z-axis in the first few hundred picoseconds (first 12 frames).
- My added proteins appear to be “frozen” in place and absolutely appear to be restrained by position restraints (they are entirely static in the the box). My solvent is also highly-ordered in the box, which tells me that the results of the equilibration step(s) is certainly a technical error of some kind since at the very least I would need to see free-moving solvent for simulation results to be valid.
I am using position restraints in my systems (for a layer of water at the bottom of my box to prevent peptide crossing), but, I was carful to outright remove the “if define POSRES” parameters in my protein’s .itp file for all of my replicates, and my solvent also has no such POSRES flag in the respective .itp. Since my protein .itp files were free of any position restraint information, and my .mdp files were also identical between systems, I have no rational explanation for such different replicate results and I’m unsure how to go about troubleshooting this apparent error in both extreme box deflation and my proteins being restrained by unknown forces.
Attached at the bottom are two photos, one of my ill-behaved replicate, and another of my well-behaved replicate which observes identical behavior as the other replicate, and is similar to other like-systems I’ve run in the past. Both systems were also minimized prior to equilibration multiple times at identical setup checkpoints, so there is no difference there (although the final minimization for the problem replicate was shorter and required ~70 less steps). Both photos are from the same final step of one of my equilibration checkpoints.
If anyone requires more information or clarification to help me with this issue, do not hesitate to ask me; I would love to have a discussion about where to go from here. Cheers.
Technical details:
I am using the martini v3.0.0 forcefields for all of my simulation assets, equilibrating with the Berendsen semi-isotropic barostat, and my pressure and temperature scaling is coupled between GroupA (protein and membrane lipids) and GroupB (water, ions, restrained water layer). My temperature coupler is v-rescale with a desired temperature of 310 K (for both groups).