It seems that you are encountering the so-called "phi-overlap" problem,
which would be quie acute when your very long c axis gets close to being
parallel to the beam. In this case, even if you are doing fine slicing as
you are doing here with 0.1 degree images, the mosaicity of your crystal may
be enough to cause those overlaps (e.g. between h,k,l and h,k,l+1) within
We had to deal with several instances of this situation in the past
couple of years. The initial images were with the long axis close to
perpendicular to the beam, hence close to parallel to the detector plane.
With those images MOSFLM started with a reasonable value for the mosaicity
(0.6 degree), but as the crystal rotated the "refined" mosaicity drifted
down to unreasonable values, ending up at 0.06 degree so that no overlap was
predicted even with the c axis along the beam! What we had to do was to
manually run spot predictions on the early images with different trial
values of the mosaicity until ALL the spots present, and ONLY the spots
present, were predicted. The optimal value of mosaicity was then given and
kept fixed, leading to the correct detection of phi-overlaps in the late
images and to the correct exclusion of the corresponding reflections from
The XDS recipe of assignment of pixels to the nearest reflection in 3D
seems clever, but I do not feel very reassured that it is a robust remedy
for dealing with this special problem. It would be very interesting to know
what you find, when reanalysing your data after your structure is solved,
about the adequacy of that procedure.
With best wishes,
On Thu, May 22, 2008 at 03:47:24PM +0100, [log in to unmask] wrote:
> Dear All
> I am processing data from a crystal for a large macromolecular complex
> with mosflm. Cell dimensions are around 120 x 150 x 650, with a p222
> spacegroup. To avoid overlaps, we have collected data with a oscillation
> of 0.1 degrees.
> When I try to process the data with mosflm, mosaicity decreases along the
> processing to a very low values (0.05 to 0.1 depending on the images). The
> result is that I miss a lot of spots from the images.
> I would appreciate any help regarding :
> 1.- What's is the best strategy to process such a dataset ?
> 2.- Which are the critical parameters in mosflm to avoid these problems,
> and how to modify them ?
> 3.- and finally, can I use this data or your advice is to try to get
> crystals in a different spacegroup ?
> Many thanks
> Best regards
> Francisco J. Enguita, Ph.D.
> Macromolecular Crystallography Laboratory
> EAN, Av. da República
> 2781-901 Oeiras
> Phone : +351-21-4469663
> Fax : +351-21-4433644
> E-mail : [log in to unmask]
* Gerard Bricogne [log in to unmask] *
* Global Phasing Ltd. *
* Sheraton House, Castle Park Tel: +44-(0)1223-353033 *
* Cambridge CB3 0AX, UK Fax: +44-(0)1223-366889 *