[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [ccp4bb]: Can I find a zinc? / weak anomalous as unbiased criterion

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

Pawel Grochulski wrote:
> ***  For details on how to be removed from this list visit the  ***
> ***    CCP4 home page http://www.dl.ac.uk/CCP/CCP4/main.html    ***
> >
> > Depends upon the quality of the crystal and your data quality and
> > redundancy. We are able to easily measure anomalous differences at the
> > Fe K-edge for a protein of similar size (soybean L-1 lipoxygenase, 94 kDa,
> > typical diffraction to 1.6 angstroms), so I wouldn't hesitate to try.
> >
> > Diana
> Hi Diana,

> I wonder how strong was your Bijvoet signal and if this would be enough
> to solve your structure doing MAD at the Fe K-edge?
> Pawel

Whether or not you have a useful signal depends not only on the ratio of 
anomalous scatterer to protein and the quality of your data, but also on
what you want to use it for. Henry Bellamy, Vladimir Shulmeister and I
collected MAD data on the chicken bc1 complex (10 Fe in 480 kDa protein
in the dimer in the asymmetric unit). We were never able to solve the
Patterson maps, so the data contributed nothing to the original phasing.
The crystals were phased instead by isomorphous replacement. Once we
knew the iron positions, we tried refining them against the MAD data
using MLPhare, but the phasing power was negligible.

HOWEVER, when we made a Bivoet difference Fourier map with the isomorphous 
phases, the Fe's stood out like sore thumbs, provided we limited resolution
to 6 A. Higher resolution contributed more noise than signal. Then as the 
phases gradually improved by multicrystal+NCS averaging and later model 
refinement, The anomalous peaks got stronger and the useful resolution
range extended higher. So if you asked me today I would say yes, we got
a really strong anomalous signal.

On the principle that any information which you don't use in refinement
can be used as an unbiased indicator of phase quality, I would recommend
using weak anomalous signals like an R-factor. This can be really useful 
in density modification procedures, where the conventional R-free doesn't 
make sense.  Your NCS correlation and R-factor will improve even if you 
have the wrong envelope and NCS operator, but if the Bivoet
difference peaks get stronger then the phases are improving.