RDKit One Component Reaction losing columns


So I am trying to carry out a reaction on a compound. I have a column that contains its ID but when the reaction takes place, the column with the ID is removed and all I have left is:


product index

reactant 1 sequence index

reactant 1

Is there a way to retain the ID?



Not sure about the output of this node (haven't had to work with reactions so far), but if you can't retain columns as you describe, you could join the resulting table with (parts) of your original table?

Thanks but there aren't any common tables to join on. None of the original columns are retained. If it just retained the ID column, it would be fine.

As even the RowID isnt retained.

I think I raised this concern in the past, it is quite tricky getting back the original data.


Thank you. I'm pleased that it's not just my ineptitude.


I think that if you use the canonical smiles rdkit node on both your reactants (i.e. input) table, and the Reactant 1 column in the output table, then you can join on those.

But, as Simon mentioned, this has definitely been flagged before, for both this and the 2 component node, that it would be nice to have an option to keep the input table rows associated with their products.


Thank you. Not sure this will work as this is part of many different nodes doing different things from a curation pipeline but I will give it a go. Also, RDKit doesn't like certain compounds and won't make SMILES from them....this I've also raised.

Have you tried generating SMILES with the Molconverter node from ChemAxon? I find it often works where RDkit fails.

Alastair I'm not sure what you mean by this?! 

Whether I use SMILES or Molfile etc, when it gets passed through the 'one component reaction', all the columns are lost, regardless, apart from the Reactant and Product columns.

Or are you suggesting the smiles as an alternative for the reacting components?



Sorry, got my threads mixed up - I think you were complaining of SMILES parsing errors.

Is there any update on this?



We will put this on our to do list. There should be indeed data available to join the reaction node's output table with the input table to take over other columns. I am planning to put in a new option to include the Reactant 1 Row ID and Reactant 2 Row ID (only for Two Componet Reactions). This option will be disabled by default so that we are backward compatible with old workflows.


Thanks for the update.