From: Theresa H. Hsu
Date: 22 February 2012 18:56
Hi all.
I have a 'learning' question based on recent thread where crystal twinning is mentioned. With the current computational methods, what types of twinnig can and cannot be solved with computers?
Thank you.
Theresa
----------
From: Andrea Thorn
Dear Theresa,
all types of twinning - merohedral, pseudo-merohedral and non-merohedral ones - can be solved given somewhat favorable conditions. In small molecule crystallography, it's quite common, especially for (pseudo) merohedral twins, it becomes increasingly popular for macromolecules as well. However, if your crystal is a split one with many domains, you should possibly look for a better crystal.
You need to treat your data correctly:
- In cases of (pseudo) merohedral twinning: Integrate in the lower - correct - space group. Not all methods of structure solution work well with twins, especially S-SAD - which is very sensitive towards data errors - and those methods that require several crystals. For refinement, the free R set has to contain also reflections related by twinning, and the random R value will go down, so don't be fooled.
- In case of a non-merohedral twin, particular attention should be paid to overlapping reflections, they should be left out at first for indexing and treated special in integration. Scaling is also more difficult. While non-merohedral twins can be dealt with, it is not (yet?) common practice for macromolecules.
Best wishes
Andrea.
Date: 22 February 2012 18:56
Hi all.
I have a 'learning' question based on recent thread where crystal twinning is mentioned. With the current computational methods, what types of twinnig can and cannot be solved with computers?
Thank you.
Theresa
----------
From: Andrea Thorn
Dear Theresa,
all types of twinning - merohedral, pseudo-merohedral and non-merohedral ones - can be solved given somewhat favorable conditions. In small molecule crystallography, it's quite common, especially for (pseudo) merohedral twins, it becomes increasingly popular for macromolecules as well. However, if your crystal is a split one with many domains, you should possibly look for a better crystal.
You need to treat your data correctly:
- In cases of (pseudo) merohedral twinning: Integrate in the lower - correct - space group. Not all methods of structure solution work well with twins, especially S-SAD - which is very sensitive towards data errors - and those methods that require several crystals. For refinement, the free R set has to contain also reflections related by twinning, and the random R value will go down, so don't be fooled.
- In case of a non-merohedral twin, particular attention should be paid to overlapping reflections, they should be left out at first for indexing and treated special in integration. Scaling is also more difficult. While non-merohedral twins can be dealt with, it is not (yet?) common practice for macromolecules.
Best wishes
Andrea.
No comments:
Post a Comment