*** For details on how to be removed from this list visit the ***
*** CCP4 home page http://www.ccp4.ac.uk ***
I have recently posted the following message to the
BB:
<< I have recently collected some datasets and most
of them suffers from the same problem: although I can
easily index individual images, the predicted spot
positions won't hold for all the dataset - I fear some
sistematic problem with the goniometer. How can I deal
with this in mosflm? >>
The problem turned out to be simpler than it looked -
after spending some time trying to index the dataset
using different starting images, I have finally
succeeded in obtaining a stable dataset processing for
all images. However, I have received some e-mails
which might be useful for other users in this
situation, and so I decided to post them here:
Petrus H Zwart (also pointed out by Tzu-Ping Ko):
<< what can happen is that the direction of the
goniometer axis is 180 degrees off. Redefining that or
assigbning a negative delta phi can solve that
problem.
if your delta phi is not to big, your data processing
program might be able to refine to the proper valuers
within a couple of images, and utterly failing
afterwards. Check the refined orientation matrix
for each image maybe. >>
James Murray:
<< There are many possible problems, and without more
information it is impossible to give a definitive
answer, but one thing I would do is check your
rotation axis. I once had a dataset given to me from
an RAXIS IV
(vertical rotation axis). Indexing on individual
images worked, cell refinement totally failed, as the
mosflm default is for a horizontal rotation axis. If
that isn't the problem, you might need to process your
dataset in small sections (say 10 images each). >>
Wei Yong:
<< Did you try to process your data set using a lower
symmetry space group? Sometimes, it causes problem. >>
Charlie Bond:
<< This is a very late reply, but in case you haven't
solved the problem, is it possible that the spindle
axis is set incorrectly, so that the software thinks
the crystal is rotating the wrong way. I have seen
this a couple of times, and what usually happens is
that as
long as the crystal is fairly close to an axis, the
refinement manages to adjust itself to fit the
pattern, but once away from the axis, there is not
enough information to do this so the refinement fails.
With denzo this is easy to test by using a delta phi
of -1 * expected delta phi (so if you collected 0.5
degree images then try -0.5). You would also notice
that the crystal rot z parameter jumps by 2* delta phi
for each successive image. Testing different spindle
settings in mosflm is probably a possibility. Another
situation where one can process a few images before
the
refinement fails is when you have an incorrect beam
centre and the data is misindexed say by one lattice
point (e.g. the program thinks hkl 001 is 000.) >>
Frank von Delft:
<< Before you go and blame your goniometer, reconsider
your spacegroup; probably you are assigning -- and
thereby enforcing -- too high geometric symmetry on
your dataset. (This is quite a common symptom, by the
way.) E.g. often things that look convincingly like
P222 from a single image are actually P2 with one
angle very close to 90. The quickest way to check is
to index with more than one image, far apart in phi.
Or just integrate as P1, and then use pointless or
xprep to judge the true spacegroup. >>
David J. Schuller:
<< I have seen similar symptoms that were indeed the
result of a systematic goniometer problem. In my case
the detector distance was off by quite a lot. Indexing
on multiple images can help in such cases. >>
Thanks to everybody,
Lucas
_______________________________________________________
Novidade no Yahoo! Mail: receba alertas de novas mensagens no seu celular.
Registre seu aparelho agora!
http://br.mobile.yahoo.com/mailalertas/
