On Mar 15 2007, Yi Xue wrote:
So far, the native crystals diffracted best to 2.4A. The MAD data
diffracted to 2.6~2.7A. We attempted to use phenix.hyss to identify copper
atoms, and the program had hard time to identify the sites.
The protein: Cu ratio is around 1:1, which is decided by ICP-AES
measurement of the crystallization sample, not derived from the number of
peaks in the anomalous map.
The crystal contains copper as a cofactor, not a soaking derivative thing.
I used phaser to do the MR, basically, what I did was to search one copy
each time, the top 20 solutions will be used to start the search of the
next copy. It stuck after finding four copies, and, I tried to change the
some paramters for searching, such as percentage for peaks, it did not
help. I would love the hear from experienced people about some tricks
using phaser to solve some difficult MR cases.
If you have even a partial molecular replacement solution, you might be
able to use the molecular replacement model as a starting substructure for
SAD phasing in the 2.0 version of Phaser. In some of our tests, this allows
us to find anomalous scatterer sites that can't be found just from the
anomalous differences, and the phases that result onced the anomalous
scatterers have been added can be significantly better than the ones from
MR alone. In your case, even if the phases weren't that much better,
knowing the Cu positions would help you to place any missing molecules.
Doing this is currently (and temporarily) a bit awkward. You have to get
the version of Phaser that comes with the current release of Phenix and
then either run jobs using shell scripts (which I could provide) or from a
beta version of the ccp4i interface (which I could also provide). The
upcoming release of CCP4 will include the newer version of Phaser that
carries out SAD phasing, plus the new interface, and a future version of
Phenix will implement this option in one of its wizards.
Randy Read