[phenixbb] Need help in refining my twinned P6322(apparent) dataset

[Phil Jeffrey]

On 3/15/11 1:18 PM, Ed Pozharski wrote:
> On Tue, 2011-03-15 at 12:12 -0400, Phil Jeffrey wrote:
>> That having been said, I want to raise the unfortunate possibility
>> that
>> you're really in P3 and doubly twinned.
>
> As in P3 with dimer sitting on the axis and resulting in apparent 6-fold
> symmetry (120+180-360=-60)?
>


In this particular case I'd envision it as potentially four 
crystallographic trimers formed by crystallographic 3-folds, giving 4 
unique monomers per asymmetric unit, with two two-fold NCS elements, 
both perpendicular to the 3-fold, and spaced 60 degrees apart (doesn't 
this beget 622?).  Actually you could probably do this with 3 monomers, 
but then you might get some apparent statistical disorder of domains 
when treating the data as either P321 or P63 where the asymmetric unit 
volume is halved.



In my case I saw something that was at first blush "P321" and pseudo 
P622, had one "good" domain and the other half of the asymmetric unit 
exhibited statistical disorder that was the superposition of two domains 
overlapping on themselves related by a twin operator.  This 
interpretation was made possible by looking at the SeMet peaks in the 
MAD map - it's not something that would be easy to come by from looking 
at a model-phased map.

I felt it was better interpreted as P3 -> P321 -> P622 doubly twinned, 
because attempts at single twin refinement and handling the statistical 
disorder did not yield a more refinable structure.  But I'm not sure how 
to prove it one way or the other.