Date: Fri, 2 Apr 2004 16:51:01 -0700 (Mountain Standard Time)
From: "Thomas E. Cheatham, III"
Subject: Re: AMBER: question about image


> I have a general question about imaging. After MD simulation in PBC, what
> kind of problems will be generated if no ptraj/image post-processing? What
> difference will be made before and after imaging? I compared RMSD of
> trajectory files before and after, and it looks no difference. Does imaging
> affect the fitting and averaging of structures? I don't really understand
> what imaging does and what kind of effects it will have.

PBC implies that the system is periodic and also that our complete system
represents 1 unit cell worth of data. However, as the system is periodic,
our particles can be in any of periodic units i.e. they can drift over
time out of the primary unit cell into adjacent. This will happen
naturally over time in MD simulation unless IWRAP=1. This leads to
the system *appearing* to blow up, when in fact if it is imaged to
put all the outside atoms back into the primary unit cell, things
look again normal.

i.e. this...

----------------
| | | |
| | | |
----------------
| |1 | |
| | xx2| |
----------------
| | | |
| | | |
----------------

...is equivalent to...

----------------
| | | |
| | | |
----------------
| |1 | |
| | x | |
----------------
| | | |
| | x | 2|
----------------

If your analysis program is set up to correctly image distances, angles,
etc, there should be no difference before and after imaging... However, if
it is NOT, the results will not be equivalent. For example, consider the
distance between 1+2 above which will be different in the two cases unless
the imaging is performed.

Molecules are typically imaged together; if you have a one molecule
psolute, then RMSd on this will not change before and after imaging. If
you have two molecules, the RMSd may change if the molecules end up in
different cells within the periodic lattice...

So, yes, in general, imaging may affect RMSd estimations and averaging of
structures as both of these commands (in carnal or ptraj) do not image.
[The distance commands in general do image, but not rms or average]

(p.s. if your trajectory file is unchanged before and after imaging then
either it didn't image OR no particles moved out of the primary unit cell
OR whatever you are RMS fitting never separated into multiple unit cells)