Explanation of the Biochemical Compound Declarative Database
Klotho takes a
high-level description of a compound's stereochemical configuration
and uses a
graph grammar
to generate a number of alternative representations (B.
Dunford-Shore and T. Kazic, in preparation).
Those
internal
to Klotho are informationally equivalent to the high-level description, while those
external to Klotho
(isomeric SMILES string and PDB file) necessarily lose significant information. However, the
isomeric SMILES
string is a common format, and has the virtue of being the
input
to
CONCORD (TM)
,
which generates a pseudo energy-minimized structure.
The
resulting PDB files are then passed to
RasMol
,
and a single image is
captured as a gif file.
A Fischer-type projection is generated from one of Klotho's internal
representations by some additional code we have developed.
Our aim in Klotho was to develop a stereochemically correct representation for
compound structures which would permit us to describe the chemical and enzymatic
mechanisms of biochemical reactions
the way biochemists do
. Since
functional descriptions necessarily entail structures, we began with those. We gradually
developed a
procedure
for checking the grammar by encoding molecules, computing the
Fischer and PDB files, and examining the high-level description and the resulting
compounds. Since we use an isomeric SMILES string as input for generating the PDB file,
we implicitly check the accuracy of the former as well. As we accumulate compounds we
have been placing their
extensional representations
on the Web.
The compounds are represented in a consistent manner, under the following assumptions.
Each compound has been checked by several different people at different times. We occasionally find errors in structural diagrams; we try to resolve these by examining crystal structures, finding more authoritative diagrams, and considering the biochemistry of reactions which synthesize or degrade the compound. We will not post dubious compounds, but occasionally find we have erred after posting.
You may be surprised by the three dimensional views: because the carbon backbone zigzags
in and out of the plane, spatial presentations will not be identical to the Fischer
projections in the plane. We have found model-building helpful in learning how to zigzag
in our heads. The substituents of a carbon below a plane running through the mean of the
carbon positions will be flipped left-right relative to the Fischer projection, while
substituents of carbons above will look like the Fischer projection. For open chain
carbohydrates, the first chiral carbon from the top should be out of the plane (and
therefore equivalent to Fischer). For the cyclic sugars, the ring oxygen is in the back.
If you haven't implemented the
Chemical MIME type
,
you can download the PDB file and view it in your
favorite viewer
,
and rotate the structures to your heart's content. If you still believe there is an
error, please don't hesitate to
contact us
.