Discussion List Archives

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Transfer from msCIF: refine_ls_class category

  • To: Multiple recipients of list <coredmg@iucr.org>
  • Subject: Re: Transfer from msCIF: refine_ls_class category
  • From: "I. David Brown" <idbrown@mcmail.cis.McMaster.CA>
  • Date: Fri, 30 Oct 1998 13:55:19 GMT
Gotzon wrote:
> 
> Nevertheless I do not agree with the separation in experimental and
> refinement classes. Perhaps you know several counterexamples, but in my
> opinion any classification of intensities at the refinement stage should
> be linked to the measurement stage. For example one could ask always how
> many reflections have been measured for a certain parity group or how many
> first order satellites. That is why I introduced _diffrn_reflns_class_code
> as the parent for the rest of codes. The same would be applicable to the
> shells. 
> 
	The problem here lies in the difference between what people ought
to do and what they want to do.  Generally cif should be permissive rather
than prescriptive.  Thus cif should make it possible for people to analyse
their refinement by class even if they have not concerned themselves with
making the measurements by class.  Indeed the measurements may not be
available if a refinement is made using structure factors taken from the
literature.  We need to make provision for this use and not require the
classes to be linked back to measurements even if, in cases such as
modulated structures, it should be encouraged or even required.  It is up
to the user (e.g. the editors of Acta Cryst.) to stipulate what
information has to be supplied.  Cif's role is to make sure the user has
the tools to report what is necessary. 

	The only places where cif rightly does not accommodate current
practice is when the current practice is wrong.  The original version of
the dictionary only allowed adp's to be given as U.  The macromolecular
community required that we added B (though the dictionary includes the
recommendation that U is preferred).  However, we have not included the
ambiguous (hence bad) beta form.  The line dividing poor (but occasionally
acceptable) and bad practice is not well defined, but forcing the user to
create measured reflection classes in order to use refined reflection
classes is beyond the mandate of cif. 

David Brown

[Send comment to list secretary]
[Reply to list (subscribers only)]