Dear Tassos,
Am 19.04.10 17:31, schrieb Anastassis Perrakis:
<snip>
The sigma issue a bit more complicated.
What we call usually sigma is the root mean square deviation (rmsd) of
the map.
Lets first recall, that the variation within the protein region is
quite large, while the solvent is rather flat.
Now, lets take an 'extreme' example, of a protein with 80% solvent.
The rmsd for that will be quite low,
since most of the AU is flat. Thus, I would argue that you might want
to consider waters in relatively 'low sigma'levels.
Of course 80% solvent will also mean that most likely this protein
will only diffract to low resolution,
so you should maybe not be putting any waters.
The inverse case argument also applies.
Similar issues, maybe more severe, come up for atom removal.
</snip>
regarding fitting electron density maps: isn't it the other way around?
Assume, that we have the same protein with the same absolute noise level
(without F000/V) in the protein region and the same loop that's missing
in the model with the same absolute density well above the protein
region's noise level, and this protein crystallizes in two crystal
forms, one with low solvent content and one with high solvent content.
Let's also assume that the solvent regions are absolutely flat. Then,
the rmsd of the total electron density in the unit cell will be lower
for the crystal form with high solvent content and higher for the
crystal form with low solvent content, as you said. However, if we want
to interprete that missing loop above the noise level in the protein
region, we would then have to contour the electron density map in the
high solvent crystal form at a higher multiple of its rmsd value than in
the low solvent crystal form to get to the same absolute electron
density level. Is this true or did I misunderstand something?
Best regards,
Dirk.
--
*******************************************************
Dirk Kostrewa
Gene Center Munich, A5.07
Department of Biochemistry
Ludwig-Maximilians-Universität München
Feodor-Lynen-Str. 25
D-81377 Munich
Germany
Phone: +49-89-2180-76845
Fax: +49-89-2180-76999
E-mail: kostr...@genzentrum.lmu.de
WWW: www.genzentrum.lmu.de
*******************************************************
