Cis/Trans info present in cycles lost in Indigo, CDK and RDKit nodes

KNIME Analytics Platform
1

Hello all,

The Molecule to X, X in (CDK; Indigo; RDKit), nodes lose information about Cis/Trans configuration in cycles: the rendering for C1C/C=C\CC1 is the same as C1C/C=C/CC1.

 

Could you confirm? Any workaround?

 

Cheers,

Fred

2

Dear Fred,

I can't speak to the other sets of nodes, but the RDKit removes cis/trans information from bonds in rings that have <8 members.

This is consistent with the fact that cyclo-octene is really the smallest ring in which an even somewhat stable trans bond can be observed.

-greg

 

3

Hello Greg,

 

Thanks a lot for your reply.

 

Unfortunately, the information is also lost for cycles that have 8, 9 or 10 members (screen cap attached).

 

Do you know why this would be the case?

 

Cheers,

Fred

4

It's only lost in the drawing (that's a limitation/bug in the drawing code). If you generate a canonical smiles for the molecule or use the "String" renderer for the RDKit molecule cell,  you'll see that the information has been retained.

-greg

5

Greg,

 

I did use the RDKit Canon SMILES node which generated the SMILES (Canonical) column that does display the correct configuration. Thanks a lot for your input.

 

I now want to export this correct configuration to an SD file.

  • The RDKit To Molecule plugged into an SD Writer node yields the incorrect representation.
  • Plugging OpenBabel out of the RDKit Canon SMILES to transform the SMILES string into an SD format also fails.

 

Any idea?

 

Cheers,

Fred

6

I believe the ChemAxon translation node is able to correctly go from SMILES -> SDF for these molecules. You might try that.

-greg

7

Yes, it's working!

 

Thanks a lot, Greg.

Fred