[gmx-users] question about g_rmsdist II

[Mark Abraham]

Fabrizio Mancinelli wrote:
> Sorry, since I hadn't any answer to my question yet, I'm resending it to
> give it visibility!

If you don't get any luck, remember that the source code is available 
for a reason :-)

> ------- Messaggio inoltrato -------
> Da: Fabrizio Mancinelli <fabrizio.mancinelli at unina2.it>
> A: gmx-users at gromacs.org
> Oggetto: question about g_rmsdist
> Data: Mon, 14 May 2007 18:52:54 +0200
> Dear all,
> 
> I would like to have some elucidation about g_rmsdist output.
> Optionally, it outputs three matrices with the options -rms, -mean and -
> scl. Questions are:
> 
> 1) Exploring .xpm files, I see there is a letter code which converts
> color exa-codes to letters to values. The "resolution" is tunable with
> option -nlevels, but the range is delimited by the highest calculated
> value. Is there a way to assign a fixed range (in order, e.g., to
> compare different trajectories of the same protein)?

There was discussion on the list of implementing a way to write a binary 
or ascii matrix. This would make it easy to do the kind of comparison I 
think you want to do. Check the archives for the resolution, but if 
there was one suitable for your needs, you'd need to use the CVS version 
of this utility.

> 2) -rms output, I guess, is the deviation from average distance, over
> time, for each atom couple. I see my matrix is full of 0's. I expected
> 0's on the principal diagonal (distance of an atom from itself is
> obviously always zero), but others? It should mean atoms' distances RMSD
> are 0, that is they do not deviate... but they don't look that fixed at
> all from the trajectory! This is even more surprisingly in the -mean
> output, which should contain, I guess, atom distances averaged over
> time. In that case there are lots of 0's too. Maybe my "I guess" are
> wrong...

You've certainly got the wrong handle on -mean. Check the man page.

> 3) What is, actually, -scl output?

Don't know.

Mark