Dear Mark and Remy,
This is what I meant by statistical disorder. It goes back to a rather cryptic remark by Zbyszek Otwinowski (Jacob, you should have googled Otwinowski instead of Ajees ;-) !) that there is a twinning continuum with sharp spots at both extremes and streaky spots in between. However, since his explanation is difficult to understand for mere mortal crystallographers like myself, I will describe below how I see it (which may not be entirely mathematically correct):
1) large twin domains: This is classical twinning. One sees diffraction of two different crystals (domains) which happen to be perfectly aligned. The crystal domains are not within coherent range and I's are added instead of F's and the spots are perfectly sharp.
2) intermediate size twin domains: I suspect the twin domains will be around 10 unit cells, but this might be completely off. Here we see streaky spots. An explanation I read somewhere is that this is due to the small domain size. In a diffraction experiment: with two slits one gets very broad maxima, with say 10 slits the maxima are already sharper and with 100 slits, one gets very sharp maxima. So in this case the domains are too small to generate sharp spots.
3) statistical disorder: Here one can no longer speak from twin domains, the unit cells (molecules?) are randomly distributed in the crystal and the crystal is "homogeneous" again resulting in sharp spots. This is what I think is the case here.
Just a 0.5 lattice translocation where the complete layer is translated will not produce the observed Patterson peak, since the intermolecular vectors are just translated and do not change. However, if a 2-fold axis is randomly skipped, two molecules which should be antiparallel are suddenly parallel and this will produce the observed Patterson peak. It could also explain the absence of alternating strong and weak reflections in the data set.
As I said, it will be a very nice puzzle to solve!
Herman
-----Ursprüngliche Nachricht-----
Von: CCP4 bulletin board [mailto:[log in to unmask]] Im Auftrag von Mark Wilson
Gesendet: Donnerstag, 3. September 2015 19:56
An: [log in to unmask]
Betreff: Re: [ccp4bb] AW: [ccp4bb] Translational NCS with one molecule in ASU
Dear Remy,
Indeed, I think you may be correct and we're pursuing this now. A perfect
0.5 lattice translocation along b in P212121 would (I think) result in the pathology we observe. We do not see zones of streaked reflections in the images, but my thinking is that if the lattice defect is coincident with a crystallographic translation operator, perhaps we wouldn't expect to.
Best regards,
Mark
Mark A. Wilson
Associate Professor
Department of Biochemistry/Redox Biology Center University of Nebraska
N118 Beadle Center
1901 Vine Street
Lincoln, NE 68588
(402) 472-3626
[log in to unmask]
On 9/3/15 12:49 PM, "Remy Loris" <[log in to unmask]> wrote:
>Dear Mark,
>
>My suspicion is that what you observe here is a lattice disorder,
>possibly related to what is described in
>
>Jimin Wang, Satwik Kamtekar, Andrea J. Berman and Thomas A. Steitz
>(2005) Correction of X-ray intensities from single crystals containing
>lattice-translocation defects Acta Cryst D61, 67-74.
>
>If your unit cell is offset statistically by 0.5 in the b-direction,
>this should provide such a strong non-origin peak as you observe. In
>the cases that have been described until now, this type of disorder
>also involves zones of nice sharp reflections and other zones with more
>streaky reflections. Do you see something similar?
>In order to use such data, they have to be corrected as described in
>the paper above.
>
>Possibly, the crystals with the 10% non-origin peak also have the
>disorder, but much less pronounced so that omitting the required
>correction did not prevent structure determination and refinement
>(similar to let say a small fraction of merohedral twinning that is
>overlooked).
>
>Remy Loris
>Vrije Universiteit Brussel and VIB
>
>DOI: 10.1107/S0907444904026721
>
>On 03/09/15 18:13, George Sheldrick wrote:
>> Dear Mark,
>>
>> Since your resolution is good enough, perhaps you should try to solve
>> it ab initio with Arcimboldo Lite. This has already solved a number
>> of structures that turned out to be unexpected.
>>
>> Best wishes, George
>>
>>
>> On 09/03/2015 05:44 PM, Mark Wilson wrote:
>>> Hi Herman,
>>> A fair point-the odds of "depressing coincidence" do seem to be
>>>climbing!
>>> We did inspect the deposited data for a similar peak and, while one
>>>is present, it is only ~10% of the origin and at a different location.
>>> We'll
>>> do some due diligence on our end by re-dissolving crystals and
>>>performing mass spec. As there seems to be some interest in this,
>>>I'll update once we've figured it out, even if it's an embarrassing
>>>case of wrong protein, same cell.
>>> Best regards,
>>> Mark
>>>
>>> Mark A. Wilson
>>> Associate Professor
>>> Department of Biochemistry/Redox Biology Center University of
>>> Nebraska
>>> N118 Beadle Center
>>> 1901 Vine Street
>>> Lincoln, NE 68588
>>> (402) 472-3626
>>> [log in to unmask]
>>>
>>>
>>>
>>>
>>>
>>>
>>> On 9/3/15 10:25 AM, "CCP4 bulletin board on behalf of
>>> [log in to unmask]"<[log in to unmask] on behalf of
>>> [log in to unmask]> wrote:
>>>
>>>> Dear Mark,
>>>>
>>>> In this case you will have to apply Baysian statistics: given the
>>>> prior:
>>>> same protein, same space group same cell dimensions and molecular
>>>> replacement fails completely, the likelihood of having some
>>>> depressing coincidence somewhere is approaches 100%!
>>>>
>>>> What I would do in addition to excellent suggestions you already
>>>>got, is to try to download the Fobs from the pdb for the structures
>>>>with the same protein, space group and cell dimensions, and
>>>>calculate pattersons with those. Sometimes strong peaks appear in
>>>>pattersons for no obvious reasons.
>>>> I would also consider statistical disorder, which will not show up
>>>>in twinning statistics since in this case F's (including phases)
>>>>are added instead of I's. Anyways, it will be an interesting puzzle
>>>>to solve!
>>>>
>>>> Good luck,
>>>> Herman
>>>>
>>>>
>>>> -----Ursprüngliche Nachricht-----
>>>> Von: CCP4 bulletin board [mailto:[log in to unmask]] Im Auftrag
>>>> von Mark Wilson
>>>> Gesendet: Donnerstag, 3. September 2015 02:06
>>>> An: [log in to unmask]
>>>> Betreff: Re: [ccp4bb] Translational NCS with one molecule in ASU
>>>>
>>>> Dear CCP4 Community,
>>>> I've had a number of helpful responses (on- and off-list) that I
>>>>will briefly summarize via response, including information that I
>>>>probably should have included in the original post. Many have
>>>>suggested a wrong space group, which I agree seems probable. MR
>>>>was attempted in PHASER with all possible choices of space group
>>>>for a primitive orthorhombic lattice, and in all cases failed with
>>>>no rotation or translation peaks above a Z-score of 5.
>>>> I've not yet tried monoclinic lattices and will, but this still
>>>>wouldn't explain (to me anyway) an apparently impossible
>>>>combination of translational NCS in P212121 with a cell that can't
>>>>accommodate a second molecule unless twinning was also present,
>>>>which may be the case (as Eleanor suggested). Others have asked
>>>>about evidence of missed weak reflections indicating a larger true
>>>>cell, which I looked for but didn't see in these images. The
>>>>crystal that was used was mounted at room temperature, so there is
>>>>no opportunity for cryo artifacts to have done something strange to
>>>>the cell.
>>>> Other suggestions included the presence of strong internal
>>>>symmetry in the molecule, which is present, but as a
>>>>pseudo-threefold, which seems incompatible with my NCS centering
>>>>operation. One respondent suggested that we've crystallized the
>>>>wrong molecule, which is something I also worried about a bit.
>>>>Although possible, the space group and cell for our crystal are
>>>>both as previously reported for this protein by another group, so
>>>>it would be a depressing coincidence if we crystallized the wrong
>>>>protein in the same cell. I'll be happy to update if/when we figure
>>>>this out should it be of interest to the board. Thank you all for
>>>>your thoughtful responses, which arrived in impressive number in
>>>>the time it took me to drive home.
>>>> Best regards,
>>>> Mark
>>>>
>>>> Mark A. Wilson
>>>> Associate Professor
>>>> Department of Biochemistry/Redox Biology Center University of
>>>> Nebraska
>>>> N118 Beadle Center
>>>> 1901 Vine Street
>>>> Lincoln, NE 68588
>>>> (402) 472-3626
>>>> [log in to unmask]
>>>>
>>>>
>>>>
>>>>
>>>>
>>>> On 9/2/15 5:30 PM, "CCP4 bulletin board on behalf of Eleanor Dodson"
>>>> <[log in to unmask] on behalf of [log in to unmask]> wrote:
>>>>
>>>>> Well - a translation of 0 0.5 0 would generate absences along b
>>>>> so that the SG could be P212121 or P 21 2 21Š
>>>>>
>>>>>
>>>>> I would suspect twinning and a monoclinic SG .
>>>>> Or as we found sadly - half the protein had disappeared in the
>>>>> crystallisation trials..
>>>>>
>>>>>
>>>>> But such a translation must mean you almost have a halved unit cell?
>>>>> Another way of saying there isn't enough room for your molecule..
>>>>>
>>>>>
>>>>>
>>>>> On 2 September 2015 at 22:38, Shane Caldwell
>>>>> <[log in to unmask]> wrote:
>>>>>
>>>>> Are you certain it's actually P212121? One possibility is you're
>>>>> at lower symmetry and the Patterson peak corresponds to the NCS
>>>>> between particles that are almost-but-not-quite
>>>>> crystallographically equivalent. In that case, MR probably
>>>>> wouldn't work. Does searching in
>>>>> P1 find anything?
>>>>>
>>>>> Shane Caldwell
>>>>>
>>>>> McGill University
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>> On Wed, Sep 2, 2015 at 5:19 PM, Mark Wilson<[log in to unmask]>
>>>>>wrote:
>>>>>
>>>>> Dear CCP4 community,
>>>>> I've encountered a curious problem with some recently collected data.
>>>>> I have a 1.8 Å resolution dataset that scales well in P212121 (log
>>>>>file available upon request). The unit cell parameters are
>>>>>similar to those reported for crystals of the same protein in a
>>>>>previous publication, although my crystallization condition is
>>>>>different. Nevertheless, my data produce a strong (47% of origin)
>>>>>peak in the Patterson map at 0.0, 0.5, 0.0, indicative of
>>>>>translational NCS. However, the unit cell parameters can
>>>>>accommodate only one molecule in the ASU without single-digit
>>>>>solvent content. Moreover, molecular replacement with a model
>>>>>that should be nearly identical fails. Standard intensity-based
>>>>>tests show no evidence of twinning or other data pathology. Any
>>>>>thoughts would be appreciated.
>>>>> Best regards,
>>>>> Mark
>>>>>
>>>>> Mark A. Wilson
>>>>> Associate Professor
>>>>> Department of Biochemistry/Redox Biology Center University of
>>>>>Nebraska
>>>>> N118 Beadle Center
>>>>> 1901 Vine Street
>>>>> Lincoln, NE 68588
>>>>> (402) 472-3626<tel:%28402%29%20472-3626> [log in to unmask]
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>
>>
>
|