Re: Simulation of a membrane protein with large extracellular domains

From: Jim Phillips (
Date: Thu Nov 13 2014 - 16:05:08 CST

Can you be more specific about it crashing "immediately"? Does that mean
on the first timestep after minimization? If you set "XSTFreq 1" you'll
get a .xst file showing the cell dimensions at every step.

My theory is that when you minimize you then start NPT at zero
temperature, which equals low pressure, so the cell shrinks. You will
probably have better luck equilibrating the system to 300K in NVT first
and then starting NPT. You can do this all in a single job as:

  LangevinPiston on
  minimize ...
  LangevinPiston off
  run ...
  LangevinPiston on
  run ...


On Thu, 13 Nov 2014, Fotis Baltoumas wrote:

> Dear NAMD list,
> I have been trying to simulate (NPT and periodic boundary conditions) the
> theoretical model of a membrane protein containing not only the
> transmembrane segment (embedded in a POPC bilayer) but also its large
> extracellular domain, covered by solvent, with no success. The initial
> energy minimization part finishes successfully, however the simulation
> then immediately crashes, giving a fatal error about the size of the
> periodic cell:
> """
> FATAL ERROR: Periodic cell has become too small for original patch grid!
> Possible solutions are to restart from a recent checkpoint,
> increase margin, or disable useFlexibleCell for liquid simulation.
> """
> Given that I have already successfully simulated the isolated
> transmembrane part of the protein using a "standard" (if there is such a
> thing) procedure for membrane simulations, I can only think that this
> crash occurs due to the large exomembrane domain of the model. I have
> tried increasing the margin and altering the values in the PBC options,
> with no success.
> The only thing I haven't tried yet is disabling useFlexibleCell which,
> however, is important for membrane simulations.
> Since in the model of the full protein the solvent& Extracelluler domain
> comprise most of the system, with the membrane/TM part being taking a
> smaller portion of the space, should I disable the flexible cell and
> simulate the system as I would do for a soluble protein? Or is there any
> other option available for this kind of problem?
> Thank you in advance,
> Fotis Baltoumas
> (P.S. I know I could just not use pressure control and be done with it,
> however I have already performed the transmembrane segment simulations
> using NPT conditions and I would like to maintain the protocol for the
> full body model as well.)

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2015 - 23:21:22 CST