CHARMM Development Project
I apologize for rather long post.I perform NPT (P=1 atm, T=298)simulation of pure POPC bilayer (64 lipids in each leaf late ,~36 waters/lipids) with CHARMM27 force field. After 7ns run area per lipid fixed to 58 A2 which is much smaller than experimentally obtained value 68 A2. I also get a longer p-p distance 46A ( expt value 39A) & high order parameter.
To expand the surface area ,ensemble was then switched to NPg(=surface tension)T (constant normal pressure =1 atm, surface tension=50 dyne/cm).
My script is
crystal define ortho 65 83 61 90 90 90 ! y axis is bilayer normal
! A B C values form coor stat of initial coordinate
crystal build cutoff 14 noper 0 ! with P1 PBC
!!-----------------image centering--------------------------------------
imag byres xcen 0.0 ycen 0.0 zcen 0.0 sele resname tip3 end
imag byres xcen 0.0 ycen 0.0 zcen 0.0 sele resname popc end
!-----------constraining H-bonds--------------
shake bonh
!---------------in/out put-----------
open read unit 30 card name "rst/po@12.rst
open write unit 31 card name "rst/po@13.rst
open write unit 32 file name "dcd/po@13.dcd
open write unit 33 file name "dvl/po@13.dvl
open write unit 34 card name "ene/po@13.ene
!-----------------dyna NPgT-----------------------
pcons pint pref 1.0 pmass 2000.0 -
surface tension 50.0 -
hoover reft 298.0 tmass 20000.0 -
imgfrq 20 cutimg 14.0 cutnb 14.0 ctofnb 12.0 ctonnb 10.0 -
cdie eps 1.0 shift vshift -
firstt 298.0 finalt 298.0 teminc 0.0 tstruct 298.0 -
iasors 1 iasvel 1 iscvel 0 ichecw 0 twindh 5.0 twindl -5.0
ewald pmew fftx 72 ffty 90 fftz 64 kappa .34 spline order 6
--------Is it correct ?
Thanks in advance.
The correct ensembles for simulating interfacial systems (such as bilayers) in CHARMM are NPAT and NPgT, with the Z axis as the bilayer normal; NPAT is usually the best choice. Also, the lattice type should be symmetric in X and Y, usually TETRagonal prism but HEXAgonal prism can also be used. For an interfacial system, NPT is really NPgT with gamma=0, which does not seem to be correct for the CHARMM parameters. I suspect that 50 dyn/cm is probably too high; our published work in this area indicates that 15-20 dyn/cm may be more appropriate for PC bilayers. The SHIFT electrostatic keyword should not be used here; fortunately, CHARMM assumes EWALD if both are specified.
Can I use simple NPT with membrane proteins in a bilayer ?

It's the same situation, with or w/o an embedded peptide or protein in a lipid bilayer; with the anisotropic pressure scaling in CHARMM, systems with planar interfaces require either NPAT or NPgT (g=gamma) ensembles, with gamma=0 being equivalent to NPT. With the current CHARMM lipid parameters, gamma=0 leads to compression of the system, while roughly gamma=15 tends to give a surface area that agrees with experiment.

The problems are that the surface tension cannot be measured directly in bilayers, and the addition of a peptide or protein would be expected to change the surface tension.
Please let me ask you more.

What if I found that the system (with embeded proteins) is not compressed too much during a few ns (or a few tens ns) ? It is eventually compressed after a long simulation ?

I'm interested in pulling a molecule to outside a membrane. In that case, NPAT is still ok or should I use NPgammaT ?

Many thanks.
The problem is that NPT is essentially NPgT with gamma=0, which may or may not be correct for your system. It is not correct for PC lipid bilayers with no added components using the CHARMM potential (or several other FFs, either). With gamma=0 we observe compression (relative to the expt area) beginning within one ns in most cases, becoming fairly compressed after tens of ns.

If the protein is large, you'll need a very large membrane, such as unit cell with an XY area of at least 6x the protein cross section for an extraction via pulling, and lots of water as well. I hope you have a lot of computer time available.
Please let me ask you again about the pulling.

Which protocol should I use for the pulling simulation ?
NPAT or NPgammaT or NPT ?

Let me say again that NPT is really NPgT with gamma=0 for systems with planar interfaces.

Assuming the CHARMM lipid FF, I'd probably try NPgT with gamma=15, so that the lipid bilayer would tend toward the expt area/lipid as the protein is removed. If you kept the area fixed, you'd be leaving a large, possibly unnatural hole in the bilayer, with no easy way for the bilayer to adjust. Your unit cell must be large enough to accommodate the expected shrinkage; the final area of the bilayer w/o protein must be at least twice the protein cross section, larger would be better.
Thanks for your kind reply.

What I'm thinking is a small size of molecule which may not perturb a membrane too much. Actually, I have seen a few papers which used NPAT for their pulling simulations (pulling out a small molecule). Do you think that this is ok or do you still recommend only NPgT ?

Many thanks.
The size of the molecule matters; a small molecule might be okay, but a peptide or protein would require more careful thought and some experimentation as to the best approach.
I see. Thanks a lot.
© CHARMM forums