Dear Xinghua,
Other have already commented on the fact that
if the spots are overlapping, even if you adjust some parameters to be
able to integrate the data and get a reasonably complete data set, the
resulting intensities will have some systematic errors that may cause
problems in downstream structure solution. One should therefore make
use of the various strategy programs that are available to try to
avoid overlap when the data are collected.
One particularly common problem is the use of a one degree oscillation
angle, regardless of the unit cell parameters, mosaic spread and
resolution. In this case, even a moderately long cell dimension of
(say) 150A that is not approximately along the rotation axis can give
rise to a large number of overlaps. There are other very good reasons
not to use a one degree oscillation angle (signal to noise) even if
there are no overlaps, so it would be to everyone's advantage if this
habit could be abandoned (Historically there were good reasons for
using one degree, but they are no longer valid). Using 0.25 degrees as
a standard would be greatly preferable.
As far as processing the images is concerned, as Harry Powell has
pointed out, increasing the "Profile Tolerance" parameters (there are
two of them, the first for the central part of the image, the second
for the outer regions) is a very good way to start. These default to
0.01, 0.01 for lab data and 0.02,0.03 for synchrotron data. It is
worth trying values of up to 0.05-0.08 instead. This will have the
effect of "shrinking" the size of the spots as far as the program is
concerned, which will automatically reduce the minimum spot
separation. It is more effective with lab data (with larger spots) and
may not work very well if you have very small spots from a synchrotron
source. I prefer this option to changing the SEPARATION parameters, if
it works.
The second thing that you can do is to reduce the mosaic spread (and
fix it at the input value). This can have a big effect on the number
of overlaps, but inevitably it will introduce systematic errors into
the intensities and this will be apparent in the FRCBIAS column in
SCALA or AIMLESS when you scale and merge the data, giving rise to
significant -ve values (ie bigger than 2-3% and negative). I do not
have any formula to decide how best to do this, but can only suggest
that you try integration with different fixed values and checking the
resulting scaling statistics (Mn(I)/sd, FRCBIAS and completeness) and
if possible try these different data sets in the downstream structure
determination to see which gives the best maps. A further option here
is to reduce the "Mosaic block size", which will effectively increase
the mosaic spread at low resolution and help offset the fact that you
have set the mosaic spread to a value that is lower than it should be,
as far as the low resolution data is concerned. You can try values
between 1 and 10, you will see immediately from the predicted pattern
what the effect is. On scaling the data, you should find that it will
help to reduce the FRCBIAS at low resolution. Again, there is no
formula for success (well, I don't have one anyway), you will need to
try different combinations of mosaic spread and mosaic block size and
see how the merging statistics (or maps) look.
Best wishes,
Andrew Leslie
On 14 May 2012, at 03:22, Xinghua Qin wrote:
Dear CCP4ers,
We collected a diffraction dataset with high percentage of spot
overlaps, It would be so kind to tell me how to ignore spot overlap
in imosflm and explain the hazard of high percentage of spot overlaps.
Thanks in advance.
Best wishes
Xinghua Qin
--
Xinghua Qin
State Key Laboratory of Plant Physiology and biochemistry
College of Biological Sciences
China Agricultural University
No.2, Yuan Ming Yuan West Road
Haidian District, Beijing, China 100193
Tel: +86-10-62732672
E-mail: [email protected]
