From: Dipankar Manna
Date: 7 March 2012 04:56
----------
From: Frank von Delft
----------
From: VAN RAAIJ , MARK JOHAN
typo correction, you'll want the long axis parallel to the rotation axis, not to the beam.
Mark
Mark J van Raaij
Date: 7 March 2012 04:56
Dear Crystallographers,
I am working on a protein having SG P6, the cell parameters are a= 79, b= 79, c= 325. The crystals are forming in big size and with very good shape. It also diffracting very well in Home source facility both in terms of resolution and intensity. But the only problem is the number of overlaps. Its showing much more than the good spots. As a result the completeness is showing maximum up to 65% even after collecting 180 degrees. I am unable to get a complete data. I tried with reducing the oscillation angel to 0.3 degree/0.5 degree but it did not improve that much. Please give me some suggestions.
Regards,
Dipankar
Dipankar Manna
From: Frank von Delft
You probably have to tilt your crystal, so that the long axis is parallel to the beam. We do this routinely: cut a plastic pipette tip to have a sharp point, then push the loop where it attaches to the pin, to bend the crystal itself.
You have to identify from your diffraction whether the long axis is pointing through the face or the edge of the loop. As it's P6, chances are it's through the face, because long-axis P6 tends to make flat hexagons which lie flush with the face. So you have to bend so the face of the loop upwards.
You'll have to practice this first, though, so put up an empty loop. Top tips:
* Don't breathe! You'll blow the cryostream away.
* Bend the loop towards (rather than away from) the rim edge of the pin to which it's glued.
* Don't breathe!
* Practise practise practise.
Another thing: most in-house sources allow you to reduce divergence of the beam. You lose intensity, but no matter, just expose longer. That also improves overlap.
Cheers
phx
You have to identify from your diffraction whether the long axis is pointing through the face or the edge of the loop. As it's P6, chances are it's through the face, because long-axis P6 tends to make flat hexagons which lie flush with the face. So you have to bend so the face of the loop upwards.
You'll have to practice this first, though, so put up an empty loop. Top tips:
* Don't breathe! You'll blow the cryostream away.
* Bend the loop towards (rather than away from) the rim edge of the pin to which it's glued.
* Don't breathe!
* Practise practise practise.
Another thing: most in-house sources allow you to reduce divergence of the beam. You lose intensity, but no matter, just expose longer. That also improves overlap.
Cheers
phx
----------
From: VAN RAAIJ , MARK JOHAN
typo correction, you'll want the long axis parallel to the rotation axis, not to the beam.
Mark
Mark J van Raaij
----------
From: Frank von Delft
Yes... quite. Thanks!
(Beam, rotation axis... how can you expect me to keep all these new-fangled inventions apart?!)
----------
From: Zhijie Li
Hampton sells an adjustable mounted loop for this purpose.
From: Jim Pflugrath
In addition to reducing the beam divergence, you may wish to use a smaller beam size by using a smaller collimator or making the slits smaller. A smaller crystal can also help to spatially separate the Bragg spots as can moving the detector closer to the crystal. Yes, closer to the crystal. This is not intuitive, but arises since modern homelab beams are not parallel but are diverging from a focal point near the sample position. It is just something else you may wish to try.
How you flashcool your sample will also have a large effect on the spot sizes/shapes.
Jim
>.....
----------
From: Frank von Delft
More accurately: "adjustable pin". Their *loops* are all adjustable (as opposed to Mitegen or MDL).
Those pins are not very useful if your crystal is already mounted. And once you've bent that pin, it can't be stored again, because robots and tongs won't close around them.
They do have the advantage of not sometimes bending back, the way bent loops do.
Those pins are not very useful if your crystal is already mounted. And once you've bent that pin, it can't be stored again, because robots and tongs won't close around them.
They do have the advantage of not sometimes bending back, the way bent loops do.
----------
From: Zhijie Li
Hi,
Besides aligning the long axis with the rotation axis, which is the most important, there are a few more things that may help:
1) Try optimizing the freezing to reduce the mosaicity (if not ideal), or shoot at RT if possible. With higher mosaicity, the shape of the reflections are elongated in the reciprocal space, increasing the chance of overlapping.
2) Collect datesets in two runs, one with detector at far position to get more spacing between dots, another one at a closer position to get high res. Sometimes you may have to sacrifice some resolution for the completeness.
3) Shoot at a synchrotron with micro beam and a large detector.
4) Try integrating the images with both HKL2000 and mosflm. My past experience seems to suggest that mosflm is more tolerant to closely positioned reflections.
Zhijie
From: Mark J van Raaij
my experiences:
C2 with a long axis parallel to the shortest crystal dimension, crystals were plates. Used prebent loops to fish the crystals. Personally I haven't tried to bend loops in mounted crystals as Frank does, but it sounds very useful.
bar-shaped P321 crystals with hexagonal cross-sections, these naturally "fell" into loops in the "right" orientations. Too "right" usually, because you also don't want the long axis too parallel to the rotation axis and have a large blind cone compromising completeness, 10-20º from parallel is great, here a minikappa or the bendable loop mounts are very useful.
Mark J van Raaij
From: Bosch, Juergen
In addition to all the excellent suggestions if you can you can also move your detector away from the center of the beam aka increase the detector size in one dimension. Not sure if you can do that at your home source though. By moving the center of the beam say to the lower 9/10 of the detector you will still have the beam position on the detector but you will also have increased the total area. Needless to say you will need to collect over a larger wedge to get a complete dataset.
Regarding your overlap issues, it does help to spend some time thinking about the experiment before collecting the data, I would like to point toward the direction of Mosflm & strategy in conjunction with the Testgen option.
Depending how badly your overlapped data is, you might want to run it through XDS and see how much it can rescue, assuming you used Mosflm or HKL2000. You can also give the Separation Close command a try in Mosflm with Postref Width and Profile Optimize (RTFM for more details).
Jürgen
......................
Jürgen Bosch
Jürgen Bosch
----------
From: Edward A. Berry
Jim Pflugrath wrote:
But the pattern is also diverging from a point at the sample?In addition to reducing the beam divergence, you may wish to use a
smaller beam size by using a smaller collimator or making the slits
smaller. A smaller crystal can also help to spatially separate the Bragg
spots as can moving the detector closer to the crystal. Yes, closer to
the crystal. This is not intuitive, but arises since modern homelab
beams are not parallel but are diverging from a focal point near the
sample position. It is just something else you may wish to try.
I'm guessing the focus point is somewhere between the crystal
and the detector, so by moving the detector closer you are better
approximating "focus on the detector" rather than "focus on the crystal"?
With a home source one probably has room for a Huber goniometer with
arcs, or better yet one of those goniometers that allows rotation
up to 90* about a point at the crystal, so the crystal doesn't move
out of the cold stream as you rotate.
One can also cheat on the mosaicity during integration by fixing
it at a small fraction of the true mosaicity. This is called
"cutting off the wings" or more euphemistically "peak height
sampling". The accuracy will suffer, but not as much as you
might expect- probably because if spot profiles are pretty
similar, the the height at peak is a good measure of peak
volume.
eab
No comments:
Post a Comment