Wilson plots are not very informative for the detection of twinning. The spikes you see in your Wilson plots, could be due to ise ring issues (both 3.89 and 2.24 A are at ice ring related d scapings.) The very large mean intensity in those resolution shells could be due to the fact that only strong reflections were not rejected during data processing. Since there is only 1 spike per dataset and the completeness in those shells is not shown, one canot be sure I guess.
The NZ plots or L-stats could be useful. Cheers Peter 2007/10/29, Iain Kerr <[EMAIL PROTECTED]>: > Thanks very much for all the suggestions so far. > > While I am pursuing all the checks and balances for twinning here are > the Wilson plots I forgot to attach before..I am not sure what is going > on, especially in B ! > > best, > Iain > > > > > > > > On Oct 25 2007, Iain Kerr wrote: > > > >> Dear all, > >> > >> I find myself posed with a rather interesting if somewhat confusing > >> problem. > >> > >> Two crystals grown from the same conditions, let's call them A and B.. > >> > >> A: > >> > >> Resolution 2.1A > >> Spacegroup P4? > >> Rmerge 0.137 (0.324) > >> Mean((I)/sd(I)) 41.0 (17.8) > >> Completeness 100 (100) > >> Multiplicity 53.6 (56.3) > >> > >> 4/mmm is clear from indexing...systematic absences show a clear 4 > >> fold screw-axis..Pointless gives the most likely as P4_1 22 (I'm not > >> clear on how it distinguishes P4_1 22 and P4_3 22..) > >> > >> Molecular replacement in Phaser, checking all the possible > >> spacegroups, gives two outstanding solutions > >> > >> LLG Z-score > >> P4_3 22 1972 41 (1mol/asu) > >> P4_3 3801 54 (2mols/asu, ASU too full warning !) > >> > >> Solutions in other spacegroups had negative LLGs and/or were rejected > >> for poor packing...the P1 solutions have LLGs of around -22000 > >> > >> I rebuilt both solutions in ARP/wARP both giving Rfree ~32% and Rfac > >> ~23%...rebuilding (most residues accounted for), adding ligands and > >> water makes no difference. > >> > >> Different iterations of refinement/rebuilding eg. cutting resolution > >> make no difference...the maps are really well defined and packing is > >> very reasonable with no clashes in either spacegroup. > >> > >> B: > >> > >> Resolution 2.3A > >> Spacegroup C222? > >> Rmerge 0.187 (0.402) > >> Mean((I)/sd(I)) 11.8 (4.8) > >> Completeness 99.4 (98.8) > >> Multiplicity 6.8 (6.6) > >> > >> Mosflm: > >> > >> 11 144 mC 255.61 64.32 63.97 90.0 90.3 76.1 C2 > >> 10 7 oC 90.69 90.74 124.09 90.3 90.7 89.7 C222,C2221 > >> 9 7 tP 63.97 64.32 124.09 90.7 90.3 90.0 > >> P4,P41,P42,P43,P422,P4212,P4122,P41212,P4222,P42212,P4322,P43212 > >> 8 5 oP 63.97 64.32 124.09 90.7 90.3 90.0 > >> P222,P2221,P21212,P212121 > >> 7 5 mP 63.97 124.09 64.32 90.7 90.0 90.3 P2,P21 > >> 6 4 mC 90.69 90.74 124.09 89.7 90.7 90.3 C2 > >> 5 4 mC 90.69 90.74 124.09 90.3 90.7 89.7 C2 > >> 4 3 mP 64.32 63.97 124.09 90.3 90.7 90.0 P2,P21 > >> 3 1 mP 64.32 63.97 124.09 90.3 90.7 90.0 P2,P21 > >> 2 0 aP 63.97 64.32 124.09 89.3 89.7 90.0 P1 > >> 1 0 aP 63.97 64.32 124.09 90.7 90.3 90.0 P1 > >> > >> This suggests pseudo-merohedral twinning to me...in C222/C222_1 ...a > >> and b are almost equivalent, so the 4/mmm symmetry would be apparent ? > >> > >> The Rmerge in 422 (19.6%) is only slightly higher than C222/C222_1 > >> ....systematic absences again suggest a 4 fold...the curves > >> calculated from the cumulative intensity distribution suggest partial > >> twinning (when inputting C222_1 into the 'old' server to calculate a > >> twin fraction from the partial twin test it says there are no twin > >> laws for that spacegroup...) > >> _ > >> The outstanding solutions in Phaser this time are: > >> > >> LLG Z-score > >> P4_3 22 1317 35 (1mol/asu) > >> C222_1 2237 46 (2mols/asu, ASU too full warning !) > >> > >> Rigid body refinement of the solutions give (C222_1 ) Rfree 43%, Rfac > >> 42% and ( P4_3 22 ) Rfree 44%, Rfac 43%....I'm thinking this is high > >> and the maps from Phaser although fitting the placed molecules have > >> poor connectivity (high Rmerge anything to do with this ?) > >> > >> Going back to crystal A it turns out the same C222/C222_1 is found > >> but lower down in the list amongst the other solutions... > >> > >> I have attached the Wilson plots for both crystals...A has a large > >> spike at high resolution (which is why I cut the data to 2.4A to try > >> and improve refinement, to no avail) and B looks horrid ! > >> > >> OK, I think that is all the information I have at the moment...have I > >> completely missed the correct symmetry..the Rmerge does seem high.. > >> > >> I have not yet tried to detwin the data (if it truly is twinned) and > >> perhaps that is impeding refinement ?? > >> > >> Any suggestions would be greatly appreciated. > >> > >> Thank you, > >> Iain > >> _ > >> > >> > >> > > > >