This is an archive copy of the IUCr web site dating from 2008. For current content please visit https://www.iucr.org.
[IUCr Home Page] [CIF Home Page] [mmCIF Home Page]

Re: Model Hierarchy in mmCIF

Lynn Teneyck (teneyckl@SDSC.EDU)
Thu, 02 Nov 95 17:06:32 -0800 

Dale Tronrud raised some good issues concerning variants in chains,
residues, and atoms.  Herb Bernstein takes issue with one of his points:

>On that point, forgive me for raising an objection to the rigid
>requirement for strict parallelism among variants.  I suspect we will
>end up tripping over that restriction.  There is no physical reason
>why multiple variants have to have identical sequences, or, even
>the same numbers of residues.  The identical cases are certainly the
>easiest models to deal with, but, if possible, it might be best
>to allow some slack for minor or even major variations in the
>representational approach right now.  

Computational or representational consistency aside, there is an
important conceptual issue.  If sequence variation, for example, is a
set of statistically independent changes, there is no harm in just
listing variant 1, 2, 3 at site X, and variants 1, 2, 3 at site Y. 
However, if there is any correlation between changes at different
sites, the strict parallelism methodology preserves this information. 
Variant 1 is then the first choice at each site, and variant two is the
second, etc.  In the example above the total number of possible cases
is three, not nine, if the changes are correlated; and nine, not three,
if they are independent.

On a more subtle level, I find Dale's example of an allele at one site
affecting the conformation of a residue at a second site compelling. 
Again, the correlation of events should be preserved.

Lynn Ten Eyck