On Mon, 2011-04-04 at 09:38 -0500, Jacob Keller wrote:
> Could it be that they are not normal because of all of the outlier,
> huge-b-factor sidechains?
That is part of it
> If every exposed sidechain without real
> density gets a b-factor of 150, wouldn't that make a sizeable and
> illegitimate non-normal population?
It will surely skew the standard deviation to higher values
> I would actually be curious about
> normality of b-factors--is there such a study/figure out there
> somewhere, with, say, histograms of b-factors of many individual
> structures?
Well, technically speaking they cannot possibly obey normal distribution
since they are always positive. Another reason is, of course, that
there are different atom types (e.g. side chains versus backbone) and
you have at best multi-modal distribution. In my experience, the
B-factor distributions always fail normality tests even when you break
atoms to groups (the most "normal" are, somewhat expectedly, waters, yet
they fail too).
Cheers,
Ed.
PS. If you keep going at this - start a new thread
--
"I'd jump in myself, if I weren't so good at whistling."
Julian, King of Lemurs
