> the Nz test says no twinning and the intensity stats say this as well.
> but the Britton and H-plots give a twin fraction of > 0.46-0.47 ! The britton and H test give an estimate of the twin fraction IF THE DATA IS WOULD BE TWINNED. The fact it gives a non zero value does not indicate the presence of twinning. > the patterson analysis suggests weak translational > pseudo-symmetry..could this be masking the detection of twinning in the Nz > test ? The peak is 5% of the origin. This is noise. > My first time using Phenix..am I interpreting these results correctly > ?..the high twin fraction would certainly explain why I get a lot of really > good MR solutions that don't refine... Of course, the fact that the intensity stats do not indicate twinning doesn't mean data is not twinned, so please go ahead in any lower space group that you/PHASER/MOLREP feels comfortable with. make sure however that the reduction you see in the R values are significant. The RvsR stat can be helpfull as well (Lebedev et al, 2005(?)). It is calculated when you supply an mtz file that contains Fcalc as well as Fobs (use obs_lables=FOBS, calc_label=FMODEL or so). Also, check your NCS operators and CA rmsd values when done with refinement to see if they suggest any higher symmetry. Peter HTH Peter
