Re: [Cdk-devel] Re: CCL: Concord/Corina replacements (With CDK)

[Geoff H. <ge...@ge...>]

On Jul 7, 2005, at 5:42 AM, Christoph Steinbeck wrote:

> You know how *normal* people (read: people who are not inspired by  
> the open source ideas) are: They try it, it fails by chance on the  
> first thing they tried, and they will never look at it again and  
> talk badly of it to others.

Maybe. Of course if that's really true, then no one would use Windows  
(after Win1.0 and Win3.x) or MS Word. :-)

This sort of question comes up again and again, and I think chemists  
are starting to look for open-source solutions. As far as the 3D  
generation goes, we have a few possible starting points, including  
PerlMol, CRK, CDK, and JOELib.

> Geoff, did you try thiophene with the version on http:// 
> www.chemistry-development-kit.org/ ?
> For me it works. Both with C1=CC=CS1 as well as with c1ccccs1.

That's exactly what I tried. Neither one works -- there are no double  
bonds, so the resulting structure has too many hydrogens and is not  
planar.

I'd be glad to cook up a "nasty SMILES" list for testing. My research  
usually generates plenty of compounds which aren't always tested well  
by cheminformatics tools. I'll see if I can present a list of SMILES  
and XYZ format.

Cheers,
-Geoff