This sounds like a case I had last year.
It looked like P41212 (in fact I solved it by MR in P43212 and refined
the data in shelxl reasonably well. Two superimposed molecules at 50%
occupancy began to emerge from the density - this was at very high
resolution).
It took me a while to get there, but in reality it turned out to be
P212121 with TWIN 0 1 0 1 0 0 0 0 -1
I generated the new asymmetric unit (2 molecules) from the old (one
molecule) and refined from there. The improvement in R-factor was
immediately obvious with both shelxl and phenix.refine.
Hope that helps,
Cheers,
Charlie
Francis E Reyes wrote:
> It seems like this space group will be the death of me.
>
> I'm working on a structure in SG P41212 one molecule per asu that was
> solved with experimental SAD phases. The resolution is to 2.5 and the
> refinement is stuck at an R/Rfree of 30 and 33 with bonds rmsd of 0.011
> and angles of 1.597 . The unit cell is 73.604 73.604 114.279 90.00
> 90.00 90.00.
>
> I'm considering the case of perfect twinning where the real s.g. is P41
> masked under the higher symmetry in P41212.
>
> It seems to be the case in perfect twinning that the approach is to
> molecular replace the refined model into the lower space group. I
> reindexed my data to the lower space group P41 and molecular replaced
> into the reindexed data with Phaser. A single was solution was found
> with 2 mol per asu (39.6% solvent content) related by NCS.
> I've refined the now two fold ncs related structure in P41 to a much
> more respectable R/Rfree of 25.2 and 28.6 with rmsd bonds at 0.004 and
> angles at 0.865 refining with a twin law and NCS as implemented by
> phenix.refine.
>
> However I'm not happy:
> [1] a simmulated anneal omit map one of the monomers in P41 where 5
> residues in a non crystal contact region of the molecule (I wanted to
> challenge the omit map) shows nearly no density. (the SA OMIT map was
> generated with phenix.autobuild using the same refine parameters as the
> final round of refinement)
> [2] the NCS selection is a little bit troubling. (maybe the phenix
> developers can chime in on this)
> reference = chain 'B' and (resseq 243:293 or resseq 310:370 )
> selection = chain 'A' and (resseq 243:293 or resseq 310:370 )
> seems as if resseq 243:293 is behaving differently than 310:370?
> [3] the densities of the side chains of a helix (not an xtal contact)
> are poorly defined, with geometry for the backbone not so good.
>
>
> There's talk about choosing the correct MR solution (see On the
> molecular-replacement problem in the presence of merohedral twinning:
> structure of the N-terminal half-molecule of human lactoferrinW. A.
> Breyer, R. L. Kingston, B. F. Anderson and E. N. Baker ) . I use phaser
> to pick my MR solution for P41. Could phaser possibly have chosen poorly?
>
> Thanks!
>
> FR
>
>
> ---------------------------------------------
> Francis Reyes M.Sc.
> 215 UCB
> University of Colorado at Boulder
>
> gpg --keyserver pgp.mit.edu --recv-keys 67BA8D5D
>
> 8AE2 F2F4 90F7 9640 28BC 686F 78FD 6669 67BA 8D5D
>
--
Charlie Bond
Professorial Fellow
University of Western Australia
School of Biomedical, Biomolecular and Chemical Sciences
M310
35 Stirling Highway
Crawley WA 6009
Australia
[log in to unmask]
+61 8 6488 4406
|