[ccp4bb] perfect twin test

[Ben Flath]

 

Hi

Firstly thanks to all who replied to my original post.

The clear consensus was to look for pseudo symmetry.

I must admit there is more to the story. Here goes the long version.

Crystals are Hexagonal bi-pyramids (under ideal conditions they are very beautiful nice crisp edges etc.  non ideal conditions crystals still grow however they lose the nice edges and look kind of like a football, and do not diffract well). Two different unit cells have been observed for these crystals; 1)  a,b=50, c=150,  2) a,b=50, c=300.  90 90 120

The data for both cells is highly anisotropic and has apparent 622 symmetry (self rotation function).  Due to the anisotropy data can only be merged to ~3 A even though there is data to ~2 A in the strong reflecting direction.

There is no pseudo symmetry detected in the small cell however there is pseudo translation detected in the big cell:

pseudo-translation vector:   0.000 0.000 0.500 (12.3%) from 'SFcheck' (what does the % mean?)

Not surprisingly intensity statistics to detect twinning are kind of all over the place but xtriage does find three twin laws (alpha for all 3 laws 0.48) and suspects the data to be twinned (in consensus with SFcheck).  Using data processed in P3 at the end of xtriage log there is this statement.

[ The results of the L-test indicate that the intensity statistics
Show more centric character than is expected for acentric data.]

I have a MR model with 45% identity.  No solutions are found in the big cell. In the small cell solutions can be found in P3212 and P32 (2 and 4 mol/ASU respectively).   Refinement stalls at ~42% and there is missing density for much of the model. I have attempted perfect twin refinement with CNS but I get huge divergence in R-Rfree  30 - 52.

SeMet protein has been crystallized but so far has only exhibited the small cell. Sites appear to have been found with SOLVE/RESOLVE and SHELX, however the maps just look like noise.

Questions:

What can cause acentric data to have centric characteristic?

Is there an option to do perfect twin refinement with phenix?

In the case of the small cell (no pseudosymmetry) perfect twin test still has values well above 2. any other ideas besides pseudo symmetry? the data also has these values in resolution shells that do not suffer form anisotropy.

All comments,questions and suggestions welcome

Sincerely

 

 

 

 

 

---------------------------------

Benjamin Flath
College of Pharmacy and Nutrition
University of Saskatchewan
320 Thorvaldson Building
110 Science Place
Saskatoon, SK
S7N 5C9