When the c axis is perpendicular to the image, i.e. parallel to the xray beam,
the only information about spacing along the c axis comes from the distance
between the lunes. There should be a circle of spots around the direct beam
with l=0, the a clear space, then a ring with l=1, then a clear space, then
a ring with l=2 and so on. As the c dimension gets longer, the layers get
closer together in reciprocal space, and these rings caused by intersection
of the l-planes with Ewald's sphere get closer together.
As the mosaicity increases, reflections whose centers are farther from ewalds
sphere will be in diffraction, and the rings get broader. In the extreme
of long c axis and/or high mosaicity the rings overlap completely and there
appears to be only one circle of spots around the direct beam center extending
out to the diffraction limit. Presumably denzo takes this to be the l=0 layer,
and since it can't find the l=1 layer assumes it is out of the picture beyond
the diffraction limit, which would be the case if you had an extremely
short c axis so the layers are far separated.
So it can get a good fit to the spot using correct a, b and a very small c.
eab
mi...@chem.ucla.edu wrote:
dengzq1987,
I am also working on a data set that has this exact alignment/mosaicity problem
noted by others. If you make a movie of your dataset that displays your images
in series, or if you just click through them quickly in adxv or your program of
choice, you will see that very few of the spots ever leave their positions on
the detector as you rotate the crystal. This demonstrates the mosaicity issue
described by Bert and Ed. In my case, Denzo seems to be able to properly
measure the a and b axes of my hexagonal cell (I know this based on a
comparison of these values to structures of homologs in the same SG), but tells
me that my c axis is only 0.21A. It seems to me that in the case of overlaps,
the dimensions of the corresponding axis would be overestimated rather than
seriously underestimated (0.21A!), so I'm not really sure of what is going on
here. Maybe someone who knows more about the software could shed some light on
this crazy measurement? I've been told by a few people that
XDS may be able to handle this dataset, but I haven't gotten to trying it yet.
I'll work on it today or tomorrow and follow up if I have any success.
Best,
Mike
----- Original Message -----
From: "Edward A. Berry"<ber...@upstate.edu>
To: CCP4BB@JISCMAIL.AC.UK
Sent: Tuesday, April 5, 2011 8:38:17 AM GMT -08:00 US/Canada Pacific
Subject: Re: [ccp4bb] how to Collecting Data from Long Unit Cell Axes ?
In the first image (spot 2) the long unit cell is obvious from the closely
spaced
spots, but this kind of image can be handled easily by Denzo. The second one
(spot 1), although at first glance the spots seem well separated, is the
problem.
As Bert pointed out the mosaicity is high, so the lunes are overlapping
(you don't see separate rings or crescents where Ewalds sphere intersects
different layers of the lattice). That means the apparently individual
spots are multiple spots one on top of the other, with the long axis
nearly perpendicular to the plane of the picture. This means there is
no information about the spacing of the long axis in this picture,and
the picture is rather insensitive to rotation, so integration is
likely to get lost unless you fix these parameters. If you fix the
parameters based on the good orientation and integrate on through,
you will be sampling the lattice rods at intervals (as the 2D electron
crystallographers say) rather than measuring discreet spots.
However if you collect while rotating around the long axis, you will
never see this orientation and should have no serious problem.
eab
dengzq1987 wrote:
hello Jürgen_Bosch,
because i have collected the data,but can't index.in some direction,the
spot is separated .but the others are set close together(picture spot1
and spot 2).so we think there is one long unit cell axes.
2011-04-05
------------------------------------------------------------------------
dengzq1987
------------------------------------------------------------------------
*发件人:* Jürgen_Bosch
*发送时间:* 2011-04-05 20:29:18
*收件人:* dengzq1987
*抄送:* CCP4BB@JISCMAIL.AC.UK
*主题:* Re: [ccp4bb] how to Collecting Data from Long Unit Cell Axes ?
What do you consider long ? 200, 300 ? 600 A ? Before shooting try to
run strategy or xplan. Move the detector back to first reliably be able
to determine your cell. Then double your estimated mosaicity and see
what strategy suggests. If you don't get many overlaps (<5%) then try a
closer distance. Don't rotate 1degrees but take 1/2 of the mosaicity.
Obviously you want to make good use of the detector area so adjust the
edges to where your crystal really diffracts. And if that resolution
leads to too many overlaps then limit your resolution and get first a
good datasets home. You then can play with 2theta for a higher
resolution dataset.
Another obvious thing to do and you don't mention what reduction program
you use is to let XDS sort your problem out. Unless you collected to
high resolution without being cautious XDS could help. If not, well then
you had your experience and now should know better.
SSRL has options to collect 450 A cells to 3A without much hassle. That
was my largest cell so far.
Jürgen
......................
Jürgen Bosch
Johns Hopkins Bloomberg School of Public Health
Department of Biochemistry& Molecular Biology
Johns Hopkins Malaria Research Institute
615 North Wolfe Street, W8708
Baltimore, MD 21205
Phone: +1-410-614-4742
Lab: +1-410-614-4894
Fax: +1-410-955-3655
http://web.mac.com/bosch_lab/
On Apr 5, 2011, at 1:05, dengzq1987<dengzq1...@gmail.com
<mailto:dengzq1...@gmail.com>> wrote:
hello all,
does anyone have the experience of Collecting Data from Long Unit Cell
Axes ? I have a crystal that diffracts to about 4 A. in some
/direction/ the spots overlap. we can't use the data to index .we
think it is because that there is a long unit cell axes. so is there
any method to solve this problem?
best wishes.
2011-04-05
------------------------------------------------------------------------
dengzq1987