I think you have just found a symmetry equivalent of your original
structure solution.
In P6122 there are 12 possible symmetry operators to choose from - only
one of which will be the identity (equivalent to alpha=beta=gamma=0)
One certainly will have alpha or gamma = 180, and beta 0, equivalent to
-x,-y,z
Then you could also have found a solution on the alternate origin (0,0,0.5)
I use lsqkab ( superpose molecule task) to fit the solution to the
original and look at the matrix to convert 1 to the other
It should match one of the symmetry operators of P6122
Eleanor
Michele Lunelli wrote:
Dear all,
I refined a protein structure in the space group P6(1)22, with one
copy in the asymmetric unit, resolution ~1.8 A, Rwork=0.20, Rfree=0.22.
Then I tried to feed Phaser (version 1.3.3) with this structure. It
found quickly a very prominent solution, but the first euler angle is
180 instead of 0 degrees (the others are 0, as well as the fractional
coordinates). This solution is not symmetry-related with the structure
that I used as search model: indeed, there are a lot of clashes.
However, when I refine this solution, I obtain immediately R factors
as good as the search model, and also the electron density map looks
perfect. Of course, I used the same reflections file to refine the
initial structure and the MR solution rotated of 180 degrees.
How can I explain this? The analysis with Truncate (moments and
cumulative intensity distribution) don't suggest any twinning, as well
as the Padilla-Yeates test. Is it possible, that I refined the
structure in the wrong space group?
Thank you in advance,
Michele Lunelli