peter Murray-rust (
Wed, 5 Jun 1996 15:17:27 +0100 (BST)

Jeffrey is quite right - this is why we started the idea of structural
formula in PPS 95. Interestingly it is much more difficult to find the
chemical structure of biologically active compounds than you might think.
And some of the standard reference works have errors.

In principle the 2-D (and stereochemically enhanced 2-D) digrams can be
created from SMILES. However the software to do this - DEPICT - is part
of a larger package and whilst this is becoming more generally available
it's still not common. There is also a tool from Synopsis which I think
is free which does this for the PC. However I believe that at this stage
it's critical to have 2-D coordinates for 2 reasons:
- the actually 2-D diagram can be drawn in different ways which,
though all correct, are of varyiong acceptability. For example 'steroids
always have the A ring at the bottom left' - algorithms don't 'know' this.
- the software isn't universal.

My medium-term suggestion is to suggest chemical/* MIME types in common
use as a way of supporting this. These should have a format which
supports 2-D coordinates. However it isn't always easy to pick the best
one and most of the software is plpatform-dependent. My own approach is
to develop Chemical Markup Language (CML) which can support all these data
types, but it's not quite ready for the hyperglossary yet.

Note that Chris Leach at Imperial College ( is
running the next electronic chemical conference and a lot of effort is
being put into trying to get authors to submit molecules electronically
(most were gifs last time).

So - at present - I suggest gifs. Make sure that any
stereochemically relevant centre is shown (wedge/hatch or cis/trans)


Peter Murray-Rust, Glaxo Research & Dev. (; (BioMOO: PeterMR)
Birkbeck College,, CBMT/Daresbury,