Re: [ccp4bb]: Disorder about a crystallographic twofold

[M.Isupov@exeter.ac.uk]

***  For details on how to be removed from this list visit the  ***
***    CCP4 home page http://www.dl.ac.uk/CCP/CCP4/main.html    ***

Dear Manfred,

I think there were some very good suggestions about your
structure and hopefully you have already sorted out your
problems. But it all sounds too familiar to me especially the bit
about the two equal solutions which can not be refined.

In our case

Structure 1996 Jul 15;4(7):801-10

Published erratum appears in Structure 1997 May 15;5(5):723

Substrate binding is required for assembly of the active
conformation of the catalytic site in Ntn
amidotransferases: evidence from the 1.8 A crystal
structure of the glutaminase domain of glucosamine
6-phosphate synthase.

Isupov MN, Obmolova G, Butterworth S, Badet-Denisot MA,
 Badet B, Polikarpov I, Littlechild JA, Teplyakov A

In the original space group P21212  R factor did not go below 38.7 %,
 and freeR below 53.1 % at 1.9 A resolution for both equal
solutions for second molecule, when the first was fixed.

However we successfully refined structure to R-factor 20.2%
freeR 27.1 %  at 1.8 A when we lowered the symmetry to P21. As we
found
out later, there were two orthorhombic sublattices in the monoclinic
unit cell, shifted by about 4 A in relation to each other.
This gave perfect pseudoorthorhombic data in monoclinic unit cell.
Rmerge was 6.2 % in monoclinic cell and 6.1 % in the orthorhombic
cell
with the same very low number of rejections.

CRYST1   70.380   82.520   86.050  90.00  90.00  90.00 P 1 21 1

The reflections in the monoclinic form followed orthorhombic
pattern so well that the R-factor of the model did not go down
and the map quality improved when we reconstituted the missing
10 % percent of monoclinic reflections from their orthorhombic
equivalents.

So I would suggest you to reprocess the data in P1, use your dimer as
 a search model and try to refine the solution in P1. This will
certainly make
density for 'crystallographic' dimer more clear. Watch out though,
the space group could still be P21, but it will be easier to check
out
after refinement in P1.

Setting occupancy of your crystallographic dimer to .5 in the wrong
space
group you effectively superimpose density for molecules at an angle
30
degrees to each other, while the non-crystallographic dimer follows
apparent C2 symmetry.

I wish you all luck with your structure
Cheers
Misha Isupov, University of Exeter