CDK vs Stereochemistry

Hi All,

I am still really new to this Knime environment and am trying to find my way around. I got my first substructure search working!

My goal is to do a stereo specific structure search, i.e. I need to be able to differentiate between, for example, endo and exo isomers. Is that possible with the CDK? Do I need another toolset for that?

Thanks so much in advance.

J

Hi J,

the substructure search is stereospecific. Playing through the endo/exo case of dicyclopentadiene, I could filter for either the endo or exo structure. There is also the SMARTS search node, which allows you to define exactly what you want.

Let me know if you encounter any issues.

Cheers,

Stephan

 

That is good news. Could you please elaborate a bit on the SMARTS node? I cannot find it in my KNIME environment, assume I need to download it?

I'll play with the 'regular' substructure node to see if I can get that to work properly, wasn't successful so far, but still learning..

 

Thanks again.

It is part of the CDK extension for KNIME, which you have already installed if you are using the Substructure node. The node is called 'SMARTS Query'. You can find it via the Node Repository.

Cheers,

Stephan

Hmmmm,

That is where I looked for it yesterday, it is not there. Is there a way to find out if I have the right version/extension of the CDK loaded?

I appreciate your help, I seem to have a hard time finding the right documentation for nodes and such.

J

Hi J,

no problem. I believe you installed the stable CDK-KNIME 2.9 branch. Until KNIME 2.10 is released, I would recommend you use the nightly build of the CDK plug-in:

http://tech.knime.org/community

It's a lot better and there you will also find the SMARTS node.

Best regards,

Stephan

P.S.: You find the version if you go to the installation wizard and check the already installed plug-ins.

Oh oh,

I downloaded what I think is what you meant, and surely now I have the SMARTS Query node. However, the two worklows that used the 'original' CDK version don't work anymore. I downloaded your example of the substructure search workflow (http://www.myexperiment.org/workflows/3342.html) and that worked fine before, but now loads with errors:

Errors loading flow variables into node : Boolean for key "charge" not found.

When I try to configure it it has a null error and will not allow to config.

My own workflow, which I mimmicked from yours, also does not work anymore, I cannot seem to get a workflow variable (as in your example) to work properly, gives error

'Invalid settings: cannot read fragment'.

I tried to feed it the fragment in a different format, did not matter.

I'll keep playing with it, but my main concern is workinjg in an environment that is unstable when a new release of nodes is downloaded.  For now I'll call it beginners 'luck'

Thanks again and regards,

J

Hi J,

the nightly is far ahead of the current stable of KNIME 2.9. I should update the myExperiment.org workflows at some point. The current nightly can be considered stable. The plug-in has undergone some major changes over the last year, hence the errors.

If you rebuild those workflows with the troublesome nodes replaced by 'fresh' nodes from the node repository, they should work.

Best regards,

Stephan

I did try that (i.e replacing the old Substructure Search nodes with the new ones) but I cannot get the fragment to be read in as a flow variable, neither in your example nor in my own workflow. Error I get:

'Invalid settings: Unable to read fragment.'

The documentation says the node expects a SMILE string, and that is what I am trying to feed it, but it won't accept it.

I must be missing something, I'll keep trying.

 

Hi J,

I have put together an example workflow:

http://www.ebi.ac.uk/~beisken/knime/

Hope thath helps.

Stephan

Thanks, Stephan. Sorry I did not get back to you earlier, but some other things came up.

Your workflow does work fine, I just found out that in order for the SubStructure Search node to work properly (even if you pass fragments through the Flow Variable port) there needs to be a fragment drawn in the configuration step. This fragment can be anything, the system seems to ignore it, but without it the node gives an error (also in your workflow). This threw me off, I thought I was doing something wrong, until I happened to draw a C-C bond in the sketcher, then it worked fine.

Thanks for all the help.

J

I see, thanks for letting me know. I have changed the node to allow for flow variable based execution without a drawn fragment. Good to hear it works now.

Stephan