Dear Chen,
based on the screenshot of CORRECT.LP, I'd guess that the space group is either [149,P312] or [151,P3(1)12] or [153,P3(2)12]. Why? Because
- Rmeas is only marginally worse than 2 times 11.0 (the lowest value), and the default MAX_FAC_RMEAS is 2 which leads to CORRECT suggesting C2
- if 149 is printed in the list, it could just as well be 151 or 153 (see http://strucbio.biologie.uni-konstanz.de/xdswiki/index.php/Space_group_determination ); you should really check the systematic absences list more towards the bottom to find out if its either 149, or 151/153 . The decision between 151 and 153 can only be made later.
Did you try all of 149 / 151 / 153 in MR? If not, that's the first thing to do.
If yes, then it could be
- either the data are not well processed, or are otherwise mediocre or damaged
- or the model is simply not good enough
Personally I find pseudo-precession images not too useful because they require the processing to have been done in P1 - otherwise you are going to impose the symmetry used during processing on the pseudo-precession image. But if you are processing in P1 then you may have incomplete data which also compromises the information content of a pseudo-precession image.
More useful, I find, is the output from pointless, in particular the part where it tests the possible symmetries and lists their Rmeas and CC values.
best wishes,
Kay
On Tue, 19 Jul 2016 15:22:44 -0400, Chen Zhao <[log in to unmask]> wrote:
>Dear crystallographers,
>
>I am now working on a 3.6 A dataset and I have some doubts on space group
>determination. Most importantly, I would like to learn and understand more
>about this procedure.
>
>The data quality is not ideal and there is some radiation damage. Base on
>CORRECT in XDS, it looks like P3_12, P3_1, C2 and P1 are all compatible
>with the reflection intensities. I attached the screenshot of the table
>here.
>
>MR is able to find a single solution (partial model occupies 63% of the
>whole) with high score in all space groups listed above. However, after a
>refmac run, only space group P1 gives a R-free below 40%, and all other
>space groups yields R-free factors about 50%. (I think I used the Rfree
>generated based on point group symmetry but I might not have done it in the
>right way.) Besides, the desired electron density (although only a very
>small region is clear) is the best in P1, present in P3_1, barely present
>in P3_12 and almost completely absent in C2. Therefore, it looks like the
>space group should be P1.
>
>However, the pseudo-procession images (attached) show clear systematic
>absences for 3_1 axis. Although to my untrained eyes, I cannot tell with
>100% confidence about the 3-fold symmetry in the reflection intensities on
>the hk0 plane, even though the lattice pattern looks like 3-fold.
>
>Twinning test does not suggest significant twinning.
>
>I just don't have much experience, but assuming it is P1, is it possible
>for NCS to have such perfect absences? Any suggestions and comments would
>be much appreciated!
>
>Sincerely,
>Chen
>
|