RE: Membrane Simulation

From: Vermaas, Joshua (Joshua.Vermaas_at_nrel.gov)
Date: Wed Oct 31 2018 - 12:49:17 CDT

Hi Alice,

What was your error with MSM again? The way you describe the system, MSM really ought to be the preferred approach, since you don't actually want your helium to interact with water if you can help it, which will eventually happen no matter how big your box is. However, since MSM is *also* incompatible with constant pressure simulations, I found I needed to add a harmonic potential to the top and bottom of my system to prevent solvent from drifting away:

tclBC on
tclBCScript {
        proc calcforces {step unique K} {
                if { $step % 20 == 0 } {
                cleardrops
        }
        while { [nextatom] } {
                set pos [getcoord]
                set z [lindex $pos 2]
                if { $z < 54 && $z > -54 } {
                        dropatom
                        continue
                }
                if { $z > 56 } {
                        addenergy [expr {$K * ($z-56) * ($z-56)}]
                        addforce [vecscale [list 0 0 -1] [expr 2.*$K*($z-56)]]
                } elseif { $z < -56 } {
                        addenergy [expr {$K * ($z+56) * ($z+56)}]
                        addforce [vecscale [list 0 0 -1] [expr 2.*$K*($z+56)]]
                }
        }
        }
}
tclBCArgs {10}

If you find that the lipids don't behave as you want, you can also try using grid forces to control the lipids.

-Josh

On 2018-10-31 11:19:15-06:00 owner-namd-l_at_ks.uiuc.edu wrote:

Hi Alice,

To deal with the layer of vacuum in your system you probably need a new
"useConstantHeight" option that is similar to useConstantArea. The
implementation of useConstantArea is pretty trivial if you search for it
in the source code, so you could potentially add this yourself. Then you
could control surface tension without worrying about the z dimension.

I suspect that if you tried to run a water-only simulation that the
surface tension of the water would collapse the periodic cell until the
vacuum was gone, so having the water-vacuum interface in your simulation
is going to provide a false measurement of the surface tension of the
membrane. I'd suggest running two membranes with water between them to
avoid this problem.

Jim

On Tue, 30 Oct 2018, Alice Zheng wrote:

&gt; Dear all,
&gt;
&gt; I am trying to simulate a membrane bilayer on top of water. In the
&gt; experiment, there is helium gas on top of the bilayer. If I use PME with
&gt; periodic boundary conditions, I have to leave vacuum on top of water. When
&gt; I use NVT for a larger system, the membrane becomes corrugated due to the
&gt; lack of a surface tension force applied to the membrane. Changing
&gt; SurfaceTensionTarget does not prevent the membrane from being corrugated.
&gt;
&gt; Is there a way to apply a surface tension force while using NVT? Or is
&gt; there a way to use NPT for the system? Are there other ensembles that are
&gt; appropriate for this type of system with bilayer lipid membrane situated
&gt; between a aqueous and a gas phase?
&gt;
&gt; I also tried using MSM and tried to eliminate the periodicity in the z
&gt; direction (perpendicular to the membrane). However, I ended up with an
&gt; error message saying that a atom is outside the MSM grid when I did energy
&gt; minimization. The periodicity in x and y could also not be displayed in VMD.
&gt;
&gt; Is it possible to have only 2D periodicity using PME? (Whenever I commented
&gt; out cellBasisVector3, I ended up with the error message that "PME requires
&gt; boundary conditions.")
&gt;
&gt; Any help would be greatly appreciated.
&gt;
&gt; Thank you,
&gt; Alice
&gt;

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2018 - 23:21:29 CST