Re: [Cdk-devel] QSAR modelling and the CDK - database

[Stefan K. <ste...@un...>]

Hi everybody,
sounds like an interesting project. Just one comment:

> > *Structures could be stored as a whole (binary or textual representat=
ion)
> > , or mimicking relational representation.

From my experience with NMRShifDB the first thing is a very bad idea. As =
soon=20
as you want to assign something (descriptors ...) to an atom or a bond, y=
ou=20
get in trouble with storing the whole structure. How to relate to a part =
of=20
it? You could use something like MDL numbers, but as soon as you change=20
something this will cause trouble with inconsistencies. So go for decompo=
sing=20
atoms/bonds in tables and do proper foreign key usage for relations.
Stefan

--=20
Stefan Kuhn M. A.
Cologne University BioInformatics Center (http://www.cubic.uni-koeln.de)
Z=C3=BClpicher Str. 47, 50674 Cologne
Tel: +49(0)221-470-7428   Fax: +49 (0) 221-470-7786
My public PGP key is available at http://pgp.mit.edu