data structure design question

[Andrew Dalke]

Konrad Hinsen wrote in message ...
> I suspect that the atom type approach is more general,
> since atoms are a much better defined concept than bonds.

>Conclusion: there is no one correct computer model for molecules.
>It all depends on what you want to do with it.

I'm at a chemical informatics conference.  They like bond
types.  They want bonds.  As far as I can tell, they've
built up their intuition with bond types.

But the general impression I get is everything is easier
to define on atoms, including properties like stereochemistry.

BTW, Thursday afternoon will be the big debut of my
PyDaylight module; see http://www.daylight.com/ for the
latest MUG meeting, MUG 2000.  I hope to wow everyone,
and get more chemists using Python!  My marketing title
is "More Science, Less Time" :)

                        Andrew
                        dalke at acm.org