Shrinking PBC box during targeted MD run

GROMACS version:
GROMACS modification: Yes(plumed v 2.6.1)

Dear users,

I executed a targeted MD run which cover 0-1ns, and its trajectory shows that the PBC box of the system is shrunken along with Z axis. This picture is taken at 1ns, you might notice that the blue line(PBC) is shrunken in Z axis , while it was initially cubic.

At that time (meaning 1ns,second picture) all thermodynamic/structural parameters are reasonable, but it collapse like this if I increase the value of constraints and run longer.

All thermodynamic/structural parameters goes chaos at this simulation.

My question is ;
1, what can be main reason to occur such shrinking/collapsing ?
2, how the result of TMD run is affected, if I continue it with the second picture ?

I’m guessing that too strong constraints may cause this kind of problem, but I’m not sure.!

Best Regards,

I assume that your protein in solvated and/or embedded in a membrane.
It may be that your system does not start from an equilibrated structure.
Did you equilibrate the system at the desired temperature and pressure?
Best regards

Dear alevilla,
Thank you for your reply.

As you assumed, this is transmembrane protein, solvated and embedded in a lipid bilayer.
This system was generated by CHARMM-GUI originally, and I followed the generated .mdp file to equilibrate it.

Here is temperature plot at last 0.5ns of NPT step.


According to this plot, I think the system is equilibrated well, at around 303K, as I desired.
Some other parameters like potential or pressure behave in the same way.


If you evaluate that the box is shrunken based only on visual inspection. Note that the picture (above) is drawn accounting for perspective. The front z line looks longer that the back z to provide the idea of deepness.
This does not mean that z-axix shrunken during the simulation. If you use the option ‘orthographic’ or similar in your visualization software to display the molecule the effect will disappear. You can also check the box definition at the end of gro file.

If the problem is not in the visualization. It can be that you have use a non-appropriate (or non-standard) setting. Usually membrane-protein simulations are performed at semi-isotropic pressure coupling or at constant volume. Under this condition the box keep is rectangular shape.

Best regards

Dear Alessandra,

As this plot suggests, it is not just a problem on visualization. Green line indicating TMD presents decrease of length of box edge along with Z, while in MD (additional simulation, not this one) it keep almost stable value.

I used semi-isotropic pressure coupling in this simulation already.

I’m guessing the problem occur because this system is not equilibrated strictly.
This is targeted MD and constraints (meaning harmonic potential) are imposed on some C alpha atoms. This potential is artificial, thus the system is not equilibrated in true sense.

According to what you said, and if my assumption is correct, the shrinking is observed because the system is not equilibrated, however it is ok since TMD simulation is not equilibrated state.

I’m not sure about this. I’d like to hear your opinion.

Best Regards,


OK, your box is still rectangular, a shrink in z-direction is coupled to an extension in x/y direction. (you use semi-isotropic pressure)

I meant to start TMD from an equilibrated structure, to exclude that what you observed is due to unequilibrated starting point. But as far as I understood you did.

What you observed during the TMD, it depends on what your target is, which force you use and on the type of system. One have to check that to use suitable steering force. Also the other system conditions should be kept (I understood you run NPT), that are P and T.
A good praxis can be to check different steering force.

I hope it helps.

Dear Alessandra,

I understand. I’ll check that how I use steering force, and try different pattern of force-time dependency.

Thank you for your kindness and advises.

Thank you again,