we have updated the nighlty build of the CDK KNIME integration to CDK 1.4.9, including some major changes in the core classes.
Amonst other things, the management of implicit/explicit hydrogens has been simplified. Hydrogens are now automatically treated as implicit and added to the compound wherever neccessary for correct property calculation. The 'Hydrogen Manipulator' node can be used to force hydrogens to be implicit, e.g., in the case where a SDFile with explicit hydrogens has been read in.
In addition, aromaticity detection and workflow building have been improved and are now more intuitive and consistent.
For any questions, comments or feature requests, please contact us.
Could you please add the search term as a flow variable in the 'Substructure Search' node in the next release? I'd like to use that node in a loop and run multiple queries against a list of structures.
I believe the following should work for you: Use the row to variable loop and, e.g., set the "Smiles" flow variable as fragment 0 in the SSSearch node. Fragment 0 is the search term used by the node.
I have attached a png showing an example workflow.
You need to create the flow variable using the 'TableRow to Variable Loop Start' node.
In the example 'sssearch.png' the Table Creator creates a column named 'Smiles'. The row elements of that column become the values of the flow variable when you execute the loop.
I had an identical workflow to your png file. Only when I explicity set the column type for the 'Smiles' column to 'smiles' did I see the additional flow variables.
On other workflows, only column type 'string' works. And yet others I can't make work at all.
Not sure why the behaviour changes, and can't see any obvious error messages in the log.
Off-line, Stephan determined that the node needs to have a dummy structure drawn in the the structure field in order for the node to recognize the presence of any flow variables.
That structure is ignored once a flow variable is selected as the input, but there needed to be something drawn in the structure field in order for the flow variables to show up in the menu.
So while Stephan works on a solution, the work around is to just draw a single C-C bond.
The node should now always allow you to set a flow variable to control the search structure without input of a dummy structure. By default, the substructure search will match everything if no input is given. That is, if the node is executed without manual configuration. This way we can make sure that the flow variable can always be set without a dummy structure.
