[Jmol-developers] Re: [Cdk-devel] 3D structure generator

[E.L. W. <eg...@sc...>]

On Friday 13 June 2003 22:59, Carl Resnikoff wrote:
> Hi Egon. That sounds like a great project. It may be too big to take on
> entirely on my own, but I would be happy to contribute to it.

Most other developers are very supporting in this regard... any question you 
may have will be answered...

> I assume for a first pass that one could either use the standard geometries
> for single, double, triple bonds between atoms of various elements, plus
> some kind of lookup table for the geometries of well known structures like
>  aromatic rings, and (as you mentioned) strained rings.

Yes, that should be possible... This would require, though, a database of 
experimental structures... 

> Or (better) one could do it entirely on the level of functional groups,
> trying to recognize each functional group in the molecule
> and then looking up its geometry in a table.
> One would also have to some kind of pattern matching to find larger
> conjugated structures like unsaturated ketones, since there geometry would
> be slightly altered, right?

Yes, that is an option... 

> Of course, chances are somebody somewhere has written algorithms
> that do some of these tasks, so a little preliminary literature search
> would probably be advised to avoid re-inventing the wheel.

Do you have access to the Encyclopedia of Computational Chemistry?
There is a chapter on "Three-dimensional Structure Generation:
Automation"

see: http://www.wiley.com/legacy/wileychi/ecc/opening.html

It references to these programs:

CONCORD: Tripos, USA, lit.rev:R.S.Pearlman, Chem.Des.Auto.News, 1987, 2 1/5-6
ALCOGEN: Chemical Concepts, no lit.rev.
Chem-X: Chemical Design Ltd, lit.rev: K.Davies and R.Upton, Tetrahedron 
Comp.Method., 1990, 3, 665-671
MOLGEO: E.V.Gordeeva, A.R.Katritzky, V.V.Shcherbukhin, and N.S.Zefirov, 
J.Chem.Inf.Comput.Sci., 1993, 33, 102-111
COBRA: Oxford Molecular Ltd, lit.revs: A.R.Leach and K.Prout, J.Comput.Chem., 
1990, 11, 1193-1205, A.R. Leach, A.S.Smellie, J.Chem.Inf.Comput.Sci., 1992, 
32, 379-385
CORINA: lit.rev: J.Gasteiger, C.Rudolph and J.Sadowski, Tetrahedron 
Comp.Method. 1990, 3, 537-547, J.Sadowski, C.Rudolph, J.Gasteiger, 
Anal.Chim.Acta, 1992, 265, 233-241
CONVERTER: Biosym/MSI, no lit.rev.

> Question: can I assume an in-memory connectivity graph of the molecule
> has already been built, or would I be parsing some compact form like Smiles
> and building the graph.

Yes, the graph is an AtomContainer, so not an efficient graph, but never the 
less a graph... it has the Atom and Bond which can be retrieved with 
getAtoms() and getBonds()... The Bond's have references to the Atoms, but
I do not think the other way around...

In the org.openscience.cdk.graph package are graph algorithms found. There is 
also subgraph searching (i.e. is X a subgraph of Y) and graph matching 
implemented...

> BTW My sourceforge login is 'studiobamboo'.

Ok, I'll add you next week... BTW, what's the story behind the name ??

Egon