I have restrained myself from commenting on Clive Delmonte's ramblings,
but now that the discussion has gotten around to a program (SHELX) that
I happen to know something about, I cannot resist rising to the bait.
First I need to correct some misconceptions. Although I believe that it
is scientifically essential to provide program sources, it really isn't
necessary for all users to read them. Indeed, because I attempt to write
highly optimized code it is not entirely trivial to understand. The
program is however provided with EXTENSIVE DOCUMENTATION!
In fact, SHELXL-97 KNOWS NOTHING ABOUT DNA; it is equally at home
refining an organometallic structure or a mineral!! It is entirely up
to the user to decide how much chemical information to supply to the
program in the form of 'restraints' in his or her input file (I have
recently put examples of such DNA and RNA restraints onto my ftp
server). In a typical DNA refinement they take the form of target
values for bond lengths and 1,3-distances, and planarity restraints on
the bases. SHELXL does not even permit restraints on torsion angles,
and it is not usual to apply any restraints between the two bases in a
base-pair, though theoretically target distances could be specified for
the hydrogen bonds. Thus this program is really NOT imposing a
particular model of DNA secondary structure.
There have been enough determinations of oligonucleotide structures by
heavy atom and more recently MAD methods that produce an (approximate)
electron density map objectively without making any assumptions about
the geometry for there to be no doubt that the reported structures are
'real' and not just computational artefacts.
To be fair, crystallographers only see structures that have
crystallized! They are fortunate that proteins have, on the whole, a
well-defined secondary structure that persisits in solution. From other
physical methods, it is clear that - except in the relatively ordered
crystal - oligonucleotides enjoy a much more varied and dynamic
existence. It is very likely that many interesting DNA motifs (some of
which may well not be based on the double helix) that arise transiently
in solution have simply not (yet) been observed in crystals.
I see no contradiction here with the crystallographic evidence: it is
totally unreasonable to come to the conclusion that the reported crystal
structures are "wrong"! They are subject to possible errors of
interpretation and experimental uncertainties just like any other
experimetal measurements, but - relative to other experimental evidence
in the biological sciences - they have a high level of reliability and
have stood the test of time very well.
George Sheldrick
PS. The current version of SHELX is 97-2, not SHELX93 as suggested in
the ramblings; for details see: http://shelx.uni-ac.gwdg.de/SHELX/
