[cctbxbb] Fail to extend space-group with centring translations

Ralf W. Grosse-Kunstleve rwgk at yahoo.com
Mon Dec 11 10:17:28 PST 2006

> I believe Ralph can handle C-1

Yes, most definitely. There are no rational rotation matrix elements
=> no problem.

> In each pair of tables 4 & 5 and 6 & 7 atomic parameters can be
> compared directly. If the P to C transformation is made, all four
> tables can be compared by inspection.

I believe you can always achieve this even with the cctbx as is. You
just may have to transform both structures to some unusual setting.
Consider that matrices with rational rotation parts elements are not
only relatively expensive to handle, but the corresponding geometry is
also highly non-intuitive. I.e. if you transform both structures to
avoid the rational rotation parts I bet the comparison is easier to
explain and understand.
The usual trick is to use space_group_type to go to reference settings,
then .niggli_cell() to go to primitive settings, then deal with the
extra translations in the primitive settings. Technically it comes
down to combining the change-of-basis matrices in the right way,
which is easy.

> In this case, the referees and editors refused to accept a non standard
> SG, so the text of the paper becomes laborious, explaining what would
> otherwise have been obvious.

If you give change-of-basis matrices from the reference setting (using
space_group_type) I think it should be easy to argue the case for
presenting the comparison using a convenient frame of reference.

Note that the space_group_type has a ".universal_hermann_mauguin_symbol()"
method to produce "nice" general space group symbols for any setting
that can be handled by the space_group class. We use this in the
automatic generation of subgroup graphs. E.g. try this:

  iotbx.explore_metric_symmetry --unit-cell "68.4 68.4 68.3 109.5 109.4 109.5" --space-group=p1 --graph graph.png

For this to work you need graphviz (open source). If you don't have it,
here is the resulting png file:


> In higer symmetry SGs it may be easier to compare structures if even more
> bizzare SG are invented, or if non-standard origins are used.

I don't think it is a problem if you use the right existing tools. :)

> Im sure this will be true, and that someone will provide the
> user-interface for generating the basis vectors form the new (probably
> enlarged) cell and new atomic parameters after the expansion.

I'd be happy to help integrate the required additions.

> Im sure the SM-tbx folk have more pressing problems to deal with than
> worrying about this minority application

The core "cctbx" (the module, not the project, see Comp. Comm.
Newsletter No. 1) is meant to be a small-molecule toolbox. Please don't
view it as "read only", unless you don't want to release your
additions as open source. (?)


Do you Yahoo!?
Everyone is raving about the all-new Yahoo! Mail beta.

More information about the cctbxbb mailing list