[gmx-users] box size for simulation of membrane protein

[Mohamed Osman]

I am going to start by quoting: Chapter 3  of the Gromacs manual (page 14)

"length of each box vector must exceed the length of the macromolecule in
the direction of that edge plus two times the cutoff radius Rc"    Then it
states:

"It is common to compormize in this respect and make the solvant layer
somewhat smaller in order to reduce the computational cost"

For a protein aligned along the z-axis in a lipid .   The protein length is
5.3 nm and the lipid layer (already equilibraited with solvants) is 6.73 nm.
  I am using a cubic box.

 I have no problem with the x and y edges  which are long to accomodate
enough lipid molecules.

My questions:
1. Protein is closer to one edge than the other.  Does this make any
difference?
   The solvants are part of equilibritaed lipid.

Along protein axis:
   a.  The periodic image is 1.43 nm away.   (along protein axis)
   b.  The distance between the protein and its mirror image is:     0.8 nm
on one side and 2.0 nm on the other side?   (These affect PME - coulmb
interactions).  Will these cause trouble?

2. Is it necessary to keep the protein at least 1.5 nm away from the edges
normal to z-axis?  This more than doubles the number of solvant
molecules.    ----

I have simulations with less than 50,000 atoms (protein, solvants, and
KcsA).  But no one realy says much about the box sizes.

What is the common practice?  Please give examples?
and
How much compromise is acceptable ?  "It is common to compormize in this
respect  !!!"

Thanks
Mohamed Osman

=======================
Professor of Electrical Engineering
Washington State University
Pullman, WA 99164-2752
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20070111/8fa96d76/attachment.html>