Is ignorant rambling, was: DNA Structure: An Age of Refinement...

[Andrew Dalke]

"Clive Delmonte" <clived at ndirect.co.uk> said:
> Moreover, any suggestion that I, or others, should address ourselves
> to such an analysis of a dozen or so different crystallographic
> refinement packages in order to spot, in maybe hundreds of lines of
> mathematics, whether the outcomes have been predetermined in any way
> (consciously or otherwise) could even give the impression that it was
> intended to be obstructive.

I assure you, it was not meant to be obstructive.  You said the
relevent literature was hard to find.  Given that, my recourse would
be to check the source code.  Sure, I haven't looked at the programs
you've mentioned, but most molecular modeling packages are less than
about 50,000 lines.  I've got access to the XPLOR 3.1 source code.
That's about 200,000 lines, but that's because it does xray and NMR
refinements and molecular dynamics and rigid body dynamics and .... 
Taking a look at what I guess is the xray refinement code, "xrefin.s",
it appears pretty cleanly written and contains 1473 lines.

> If someone gave me a text in Arabic, Hebrew, Cuneiform, Greek, etc.,
> even with a dictionary,

  Irrelevant analogue for three reasons:
1) If you want to do serious study of the Bible, you learn Hebrew and
Greek.  Same is true with software.  If you don't understand the source,
it would be hard to take you seriously.
2) Given time you can learn those langauges.  Obviously graduate
students, even those who have not "done a lot of programming in
FORTRAN", are able to work on xtal refinement programs in less than a
year while taking classes and starting their own research.
3) If you believe there is an unconscious predeterminism towards
a double helix structure, you are unlikely to find that assertion in
published statements, so you must examine the source code.

You must have spent quite some time acquiring the references you've
listed.  You could have spent comparable time examining the source
code to verify your assertions.  Plus, after you understand one such
program, the rest are much easier.

> I'd be just as totally lost as having to follow the logic behind
> a complicated crystallographic package in source code which has
> been developed and elaborated over many years.

Then ask for an early version of the source code, like SHELX-76.
Most people keep older versions around in case they need to reproduce
past results.

Besides, your original statement was:
> but, unfortunately, explicit details of the working
> of the algorithms, and the context of their use, do not seem to be
> widely and publicly available.

when your complaint now is not that the details are not available, but
that the details are not available in a sufficiently simple form that
*you* can understand them.  Apparently they are in a form usable and
understandable by most people in the field.

  Given that you have only a non-expert's understanding of
crystallography, how are you sure that your interpretation of the
various references isn't skewed by your inexperience?


> In every other field of biophysics and biochemistry, researchers
> usually lay out each main stage of their work in explicit,
> comprehensible experimental detail, and explain the philosophical
> development underlying their work, as well as the way they carried it
> out.

  I beg to differ.  The xtal (and NMR) people I've met have just the
same practices as others in biophysics and biochemistry.  After all,
there are people who practice in several subfields; why would they
change their habits that drastically?  I find it much more likely that
you have misinterpreted your research than that such a blanket statement
is true.

> It is my experience, though perhaps not yours, that explicit statements
> of the philosophical approach being adopted towards the prior
> acceptance of plectonaemic winding in DNA in an algorithm is extremely
> rare.

  I've already told you I'm not a refinement person.  I can tell you
that the molecular dynamics force field, which is based on small molecule
experiments and hence has no bias for the specific DNA conformation,
gives reasonable energies for a double helix and so that conformation
should be expected quite often, theromodynamically speaking.

  This is my background, and it is the reason I believe that a double
helix model for (most) DNA is correct.  The only way to convince me
otherwise is to come up with a model with better binding energies in
a standard salt/water environment.  You've not done so.

> Could it be that the crystallographic outcomes are not actually as
> secure as the strident claims made for them might induce us to believe ?

  You obviously have a hard struggle ahead of you to prove your thesis.
Since the others who have read the literature diagree with you, you will
have to be explicit in showing how things are different.  One way is to
come up with a better DNA model, another is to delve in the source code
and show how an implicit double helix model is built in and overly
prejudices the results.

  You've not done so, and I predict that until you do you will continue
to be frustrated in your attempts to show the validity of an alternate
model.

> It would be so unfortunate if any ferocious and unhelpful responses from
> oligonucleotide crystallographers came to be seen as deliberately
> intimidatory, wouldn't it ?

  It would also be unfortunate if you interpret all counter-claims by
"oligonucleotide crystallographers" as ferocious and unhelpful.  What
if they are right and DNA most often is a double helix?


						Andrew
						dalke at acm.org