STARANISO anisotropy & Bayesian estimation server.

The STARANISO Server

Anisotropy of the Diffraction Limit
and
Bayesian Estimation of Structure Amplitudes

Advice on inclusion of test-set (Rfree) flags

We are grateful to Andrew Leslie (MRC-LMB) for identifying a potentially undesirable 'feature' of CCP4/REFMAC (and possibly other refinement software), namely the default 'fill-in' of unmeasured ('missing') data outside the diffraction cut-off surface with DFcalc. In general, any treatment of STARANISO's output data that adds missing reflections out to the diffraction limit may result in a very high proportion of unobserved data compared with data that have had a cut-off applied according to an isotropic criterion. Where missing reflections have been added in this way the user should take care to verify that any applications used to post-process the data can handle it correctly.
In order to deal with this problem, an extra column (label = 'SA_flag', type = 'I') has been added to the STARANISO output MTZ file. SA_flag has one of 4 possible values, as follows:

SA_flag value Point color Meaning*
0 or MNF (missing) Grey Unobservable
1 Dark blue Observable
2 Red or pink Unobserved
3 Orange through white Observed

*See the GLOSSARY for an explanation of the terminology used here.
Note that currently only values 1 and 3 are output and a new checkbox has been added to the data entry page to use this flag. If left checked the uniqueify script is used to add test-set (R_free) flags to the output merged MTZ file from STARANISO out to the isotropic diffraction limit. Then the file is copied (using SFTOOLS), but omitting 'unobserved' and 'unobservable' reflections outside the anisotropic diffraction cut-off surface where the SA_flag has a 'missing' value. Finally, the SA_flag column is removed entirely (to avoid the possibility of confusing some refinement programs). The BUSTER structure refinement program (Global Phasing) is 'SA_flag aware' so the SFTOOLS steps are not actually necessary (but harmless) provided that BUSTER is the only refinement program that will be used.
Provided the box for adding test-set (R_free) flags is left checked these steps are now performed by the server as described above (unless test-set flags were already present in the merged input data) and therefore additional user action is no longer required.
If the box for adding test-set (R_free) flags is unchecked, the behaviour of the software is as for the previous version and therefore addition of test-set flags becomes the responsibility of the user (again unless test-set flags were already present in the merged input data). This will be the case where the user wishes to add test-set flags from a reference dataset. Unless only the BUSTER refinement program will be used, this must then be followed by removal of reflections with missing SA_flag value, and (where necessary for the refinement program) removal of the entire SA_flag column. For this purpose the following script can be used (the identical script is used on the server if the option is selected):
sftools
> read input.mtz
> select col SA_flag present
> delete col SA_flag
> write output.mtz
> end
The result of this is that REFMAC (or other refinement software) will only fill in the 'observable' missing reflections (blue points) with DF_calc that lie inside the anisotropic diffraction cut-off surface, which is correct, and not to the 'unobservable' missing reflections (grey points) and 'unobserved' (red or pink) that lie outside the cut-off surface.

SA_flag value	Point color	Meaning*
0 or MNF (missing)	Grey	Unobservable
1	Dark blue	Observable
2	Red or pink	Unobserved
3	Orange through white	Observed

The STARANISO Server

Anisotropy of the Diffraction Limitand Bayesian Estimation of Structure Amplitudes

Advice on inclusion of test-set (Rfree) flags

Anisotropy of the Diffraction Limit
and
Bayesian Estimation of Structure Amplitudes