[gmx-users] Decreasing of the box size during membrane protein simulation

[James Starlight]

Dear Gromacs Users!

I'm simulating membrane receptor embedded in the explicit membrane
using charmm36 ff.
My simulation setup consist of

; Compound        #mols
Protein             1
iso                 1
POPC              228
SOL         11713
NA               66
CL               70


I've done short nvt and npt equilibration using sd1 integrator as the
thermostat and Parrinello-Rahman barostat
; Temperature coupling
; tcoupl      = no                     ; modified Berendsen thermostat
tc-grps     = Protein_LIG POPC SOL_Na_Cl
tau_t       = 2   2 2                    ; time constant, in ps
ref_t       = 310   310 310                    ; reference
temperature, one for each group, in K
; Pressure coupling
pcoupl      = Parrinello-Rahman             ; pressure coupling is on for NPT
pcoupltype      = semiisotropic
tau_p           = 5
ref_p           = 1.01325 1.01325
compressibility = 4.5e-5 4.5e-5
refcoord_scaling    = com

during analysis of the 50ns production run trajectory I noticed
significant decrease of my box size on Z dimension (looks like my
system was compressed): from 10 to 6 nm. How I could avoid such
compression ?


Thanks for help

James