On Thu, 2014-01-09 at 18:03 +0900, kurisaki wrote:
> Dear Amber developer,
>
> I find that "image" command with "familiar" option are prepared
> for a system in truncated octahedral box.
> However, I cannot find any difference between with-"familiar"
> and without-"familiar" when I analyzed a system in rectangular box.
> (I am analyzing water molecules around protein, thus "image"
> should influence the following analysis.)
>
> Please assume that we analyze a system in rectanglar box.
> If we use "image" command with "familiar" option for the system,
> does it cause serious artifacts such like atomic crash of water
> surrounding protein, basically?
> Or does ptraj automatically neglect "familiar" option
> when it treat a system in rectangular box.
The answer is that a "familiar" representation of an orthorhombic box
(with all 90-degree angles) is equivalent to standard, non-familiar
imaging, so there appears to be no difference.
All of the standard periodic shapes we use (orthorhombic, truncated
octahedron, rombic dodecahedron, etc.) can be rearranged as a general
triclinic cell defined by 3 vectors and the angles between them. The
standard imaging approach rebuilds this general triclinic cell, so any
non-orthorhombic cell (like a truncated octahedron) will look quite
different than you would expect it to. When you use the 'familiar'
keyword, cpptraj and ptraj will automatically take the most spherically
symmetric representation of that unit cell (i.e., they will take the
image of all atoms that are closest to a pre-defined 'origin'). When
all angles are 90 degrees, the standard triclinic cell _is_ the most
spherically symmetric representation---this is why 'familiar' and
non-'familiar' are equivalent in this case.
HTH,
Jason
P.S. As an aside, cpptraj has a nifty "autoimage" command that will
automatically do the 'centering' and 'imaging' for you, invoking
'familiar' when need be. It is quite helpful in cases like these.
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Jan 09 2014 - 08:00:03 PST
