***  For details on how to be removed from this list visit the  ***
***          CCP4 home page http://www.ccp4.ac.uk         ***


> In any case, one could always refine a continuous sequence with a 
> contiguous numbering and, at the end of the process, relabel/renumber

Renumbering consecutively leads to loss of information, which cannot
be recovered w/o keeping extraneous records. I think that should be
avoided. I see no technical problem distinguishing between a sequence GAP
and a real LINK (be it cis or trans, but we can default to TRANS with
little risk). As IT said before, a reasonable distance cutoff seems
feasible. Perhaps 95% of all cases could be distinctly distinguished.

The TER record might seem as a means to indicate gaps in the file,
but that leads to another problem: some check programs (haven't deposited
lately, maybe PDB can comment) complain if OXT is missing at the
terminal record. However, putting an OXT on a chain that
continues and requires a N+1, is wrong. This decision requires a 
cross-check against sequence. Still, it is doable.   

Where all this becomes relevant I believe is not at the crystallographer's
level
(of whom we presume is knowing what he does) but later at the USE of
the structure. Many of these nuisance or housekeeping errors need
to be fixed for ligand docking, and I remain unconvinced that all
the 'fixing' by these programs such as flipping residues, adding 
missing parts, etc) without re-refinement are entirely defensible.

Nevertheless, I see no excuse for housekeeping errors and nuisance 
such as obvious NQH flips, stray waters, etc. This should be fixed before 
deposition. It sucks, I know, but better tools will help. I think we have
already pretty good programs, and user input should really be an
encouragement to the programmers. We like these programs, although
at 3 am it is occasionally difficult to suppress a curse.

Suggest we close on that note.

Best, br 


Reply via email to