Re: Amore correlation coefficients

[Stephen Soisson <soissos@rockvax.rockefeller.edu>]

***  For details on how to be removed from this list visit the  ***
***    CCP4 home page http://www.dl.ac.uk/CCP/CCP4/main.html    ***

Thanks to all who have responded to the question about funky correlation
coefficients.

My data set only extends to 4 Angstroms resolution, so it is somewhat
difficult to estimate the overall B for the data from a Wilson plot. 
But, it seems to be somewhat high, probably about 50 A^2 .  This is
supported by the fact that the data are very strong to about 7 Ang. and
then fall off in intensity quite steeply as you extend to 4 Ang.  I
know, this is not the best situation to be in, but we take what we can
get.  This type of data was observed in the K+-channel structure, and
was correlated with high overall B's for the data and the model.

As for the search model, The average B is 44 A^2.  This is primarily due
to some loop regions that have high B's.  I decided to leave the
B-factors as-is since this would, in effect, down-weight the flexible
regions relative to the rest of the structure.  Does this sound
reasonable?

Any thoughts on how this all relates to the correlation coefficient
business would be great.

Many thanks-

Steve


Eleanor J. Dodson wrote:

> > In a recent MR case using Amore, I noticed the following phenomena:
> >
> > Using my straight reflection data, the correlation coefficients observed
> > during rotation and translation steps were relatively high (30%) which
> > seems peculiar since I was searching with only 1/4 of the a.u.  When I
> > put the solutions into the rigid body fitting function in Amore, it
> > re-calculates a scale and B-factor for the data, and the STARTING
> > correlation coefficients are much lower now, (22%).  Meanwhile, the
> > fitting step is applying a B-factor of -90 to my data.
> >
> > I have used AMore many times in the past, and never seen these sorts of
> > problems.
> 
>  What are the average B factors for the model, and for your data? what
> is the resolution and completeness of data? all these are relevent.
> 
>  Eleanor

-- 
------------------------------------------------------------------
Stephen M. Soisson, PhD                Lab of Molecular Biophysics
soissos@rockvax.rockefeller.edu        Rockefeller University
Phone: (212) 327-7915                  New York, New York  10021