Molecule to CDK loses aromaticity.

We recently started using a molecule collection containing aromatic SMILES with lower-case atom labels: c1c[nH][n]2cccc[n]2[nH]1

When these are passed through the Molecule to CDK node, the resulting structures are missing all double bonds, or even normalised rings, displaying fully saturated systems (the Molecular Properties node gives 0 aromatic bonds and atoms).

A work around is to use the OpenBabel node to convert SMILES to SDF first, but then we need to install OpenBabel for all our users before we distribute the workflow.

Are we missing a simple solution?

KNIME 2.3.1 (Both Win & Mac)
Molecule to CDK settings:
Replace column
Generate 2D coordinates
Force generation
Add explicit hydrogens
Processing timeout = 10,000


That is very likely a problem in the CDK library. However, we are using a quite old version (currently simply lack of time to change to the latest one), so this one might be already fixed.

Any timeline on a CDK update? 2.3.2? wink

Another workaround in the meantime is to get our users to load all the community nodes and use the 'Molecule to RDKit' node just for this particular workflow. Many of the other nodes expect CDK input, so we won't be using that as a general replacement.




An update: this problem seems to be solved in the current version of CDK (