Re: [Cdk-devel] CDK aromaticity detection

[Egon W. <ego...@gm...>]

On Wed, Mar 26, 2008 at 3:34 PM, Christoph Steinbeck
<ste...@eb...> wrote:
>  I was wondering about the state of the CDK aromaticity detection.
>  The AllRingsFinder seems still be play a role in the code, which it
>  ideally should not. I thought I remembered that someone was really
>  working on this to remove it.

I wrote a new tool:

http://cdk.svn.sourceforge.net/viewvc/cdk/cdk/trunk/src/main/org/openscience/cdk/aromaticity/CDKHueckelAromaticityDetector.java?view=log

The AllRingsFinder *is* involved, but it is only applied to systems
with up to three rings... So, the combinatorial explosion is rather
minimal :)

>  Trying to use the CDK fingerprinter on one of our databases still gives
>  us quite a bit of AllRingsFinder timeouts.

Not good... :(

Really, not good... because this would mean that it is damn slow even
for three (connected) rings, or less!

Can you email me (offline?) test code with input structures?

Egon



-- 
----
http://chem-bla-ics.blogspot.com/