Dear Eleanor -

That is correct.  The pseudo-sg is P6, and the structure has been  
refined in this sg.  The intensity difference between the strong and  
weak subsets is quite significant that for most data sets, auto- 
indexing routines will miss the weak spots and pick the pseudo-sg  
instead.  The pseudo-sg is a'=b'=90, c'=56, the true sg is a=b=156,  
c=56.  Note that a = a' * sqrt(3).  So, the sg assignment is certain.

Owen

On May 20, 2010, at 6:33 AM, Eleanor Dodson wrote:

>   This looks a bit strange..
>  If you have a hexamer in the asymmetric unit, in P3, then that means
> all symmetry copies lie in the same plane. To generate the Patterson
> peak, 2/3,1/3,0 the hexamer must be centred at 1/3,1/3, z
> (with symmetry equivalents  0,-1/3,z and -1/3,0,z )
>
> I would expect ypu to have a pseudo higher symmetry SG - does  
> pointless
> make any suggestions?
> Eleanor
> Jürgen Bosch wrote:
>> Hi Owen,
>>
>> you should also make the following plot with your data:
>> y-axis relative intensity of off-origin peak versus x-axis  
>> resolution cutoff used for calculation (30 Å - 4 Å in 2 Å steps).
>>
>> You can have multiple cases of shifts and I would start with a  
>> perfect hexamer first, take some random monomer and apply a perfect  
>> sixfold, move it along the axis where it should be in your crystal  
>> lattice (things get more complicated if you have a top/down  
>> hexamer, so keep it simple). Now if you shift your hexamer to  
>> 2/3,1/3,0 your plot should yield a straight line and be independent  
>> of resolution. Now start rotating the second hexamer relative to  
>> the first clockwise with your sixfold, I would use increments of 3  
>> degrees, which will result in 19 models, then recalculate the off- 
>> origin peak heights and see if they match up with your data. I  
>> should note  here, if your real data does not show a strong drop in  
>> peak height of the off-origin peak, then you most likely don't have  
>> a slight rotational translation in your second hexamer.
>>
>> One other important thing you should look at is the relative  
>> orientation of your sixfold axis, is it truly perfectly aligned  
>> with one of the cell axis ? If not fix this in your model,  
>> otherwise your calculations will be of academic nature. For this  
>> particular case the use of GLRF is more helpful than MOLREP (sorry  
>> Garib, but maybe Garib can come up with a solution to zoom into  
>> certain peaks like you can do in GLRF).
>>
>> When the tilt is fixed you should be able to figure out the  
>> rotational translation in your second hexamer.
>>
>> Enjoy your puzzle,
>>
>> Jürgen
>>
>> P.S. P3 is certain ? Check with pointless or by human brain visual  
>> inspection (HBVI)
>>
>> On May 15, 2010, at 11:53 PM, Owen Pornillos wrote:
>>
>>> Dear ccp4bb –
>>>
>>> I have questions with regards to crystal disorder that gives rise to
>>> translational pseudosymmetry.
>>>
>>> We have a rotationally hexameric protein that crystallized in P3,  
>>> with
>>> one hexamer in the asu.  The local 6-fold axis of the hexamer is  
>>> non-
>>> crystallographic, and is essentially parallel to the  
>>> crystallographic
>>> 3-fold, which gave rise to translational pseudosymmetry.   
>>> Intensities
>>> for the (h,h+/-3n,l) reflections were on average about 8 times
>>> stronger than the weak reflections, and the native patterson gave an
>>> off-origin peak about 70-80% of origin (depending on the crystal) at
>>> fractional coordinates (2/3,1/3,0).  We are hypothesizing that the
>>> break in local 6-fold symmetry is caused by small rigid-body
>>> displacements of each subunit (as opposed to conformational  
>>> changes in
>>> the protein), and we are trying to estimate the magnitude of the
>>> displacements in the crystal.
>>>
>>> To do this, a perfectly symmetric hexamer with the local 6-fold axis
>>> parallel to the crystallographic 3-fold was generated, and then  
>>> shifts
>>> were introduced to the atomic coordinates.  The direction of the  
>>> shift
>>> was chosen randomly for each atom, and a single magnitude applied to
>>> all atoms, which was then changed incrementally.  Structure factors
>>> were calculated from these models, and their pattersons were
>>> examined.  The magnitude of the off-origin peak could be reproduced
>>> with an atomic shift of say, 1 Å.  Because all of these calculations
>>> were made with synthetic structure factors, this is not  
>>> necessarily a
>>> reliable estimate.  The questions are, how far off are we, and in  
>>> what
>>> direction (i.e., are these shifts underestimates or overestimates)?
>>> Is there a way to obtain a reliable estimate?
>>>
>>> Thanks in advance,
>>>
>>> Owen
>>
>> -
>> 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/
>>
>>