[ccp4bb]: Summary of Replies: Data Reduction for mosaic crystals...

[raji <raji@lamar.colostate.edu>]

***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***

Hi,
Here are all the responses to my questions! Thank you very much.
Raji

Questions:
I am working on a dataset where the crystal has a very high mosaicity - my 
estimate is ~2.0 or more. I have had to take 0.1deg frames during data 
collection. Denzo V. 1.97.2 does not seem to be scaling the data very
well.  Would anyone know if HKL2000 can do a good job or if there's any other 
package thats good in dealing with minislices. Also, any suggestions about how 
to treat and process minislices?
Replies:
Several people suggested the same:
We've processed data with slices of 0.2 deg with MOSFLM without problems.
         > A reference on super-fine phi-sliced diffraction:
         > J. Appl. Cryst. (2000). 33, 1187-1188
         > J. Lovelace, E. H. Snell, M. Pokross, A. S. Arvai, C. Nielsen, 
N.-H. Xuong, H. D. Bellamy and G. E. O. Borgstahl
The author of d*Trek (Jim Pflugrath) has written a very readable paper on this 
topic (from his talk at a CCP4 Study Weekend); Acta Cryst. (1999). D55, 
1718-1725: 'The finer things in X-ray diffraction data collection'.

Seems to me like a typical case where you should use XDS... well, it seems 
more like like a typical case of bad data collection strategy ... ;-) It is in 
general pointless to collect phi slices that are less than half the mosaicity.
Why collect thin slices with high mosaicity ? Fine phi slicing gets really 
powerfull with low mosaicity crystals only! Typically people reduce the phi 
rotation in order to avoid overlaps. But, reducing it significantly lower than 
the mosaicity is pointless. ie. The reflections will be overlapping simply 
because they do overlap since they are much too 'wide' ... Thus, fine
slicing will not help solve the problem and it will add very little to data 
quality when mosaicity is high.
The best choice is to grow slightly different crystals, freeze them better, or 
NOT freeze them  (people tend to forget that you can actually solve a 
structure without cryo-freezing ...) Steve Gamblin and Co had once good luck 
with VERY mosaic crystals in the microfocus (>2.5), shooting just the edge of 
the pillow-like crystal and I am told that this was the only thing that worked 
for them. But try the simpler things first.

Several people gave the same suggesion: To my knowledge, the best software in 
this case is XDS, from W. Kabsh.

Some people suggested: You might also want to try D*TREK. It handles fine 
sliced data without any trouble usually however if your crystals really are 
that mosaic you may run into other difficulties resulting from overlapping 
diffraction peaks etc. no matter how finely you slice!

Here are a couple of tricks that may be adequate if you don't want to spend 
time learning other programs:
Sometimes scalepack can be unstable with fine phi slices because the 
'reference frame' only contains very small partials. I found that if I tell it 
to fix the scales of 3 or 4 consecutive frames then this gives a better 
reference. If things are too bad for that you can scale the data to a 
pre-scaled 'reference' data set - if you have one, and then remove the 
reference data and read in the scales. The reference data doesn't have to 
agree with the new data very well, but make sure the sigmas in the reference 
data represent the errors between the reference and new data.

With such a large mosaicity and such thin slices you have to use tight 
restraints in SCALEPACK. scale restrain 0.01 and b restrain 0.5 or something 
like that.

Rajeswari Edayathumangalam
Dept. of Biochemistry & Molecular Biology
Colorado State University
Fort Collins, CO 80523 USA