Hi guys,
I’ve encountered an issue while attempting to standardise the orientation of molecules with a common scaffold using KNIME. This process is crucial for facilitating the visual comparison of molecular structures.
To achieve this, I used the RDKit Generate Coords node, configured to generate 2D coordinates, and employed the Maximum Common Substructure (MCS) of the compounds of interest as a template SMARTS. Generally, this method successfully aligns (i.e., rotates) molecules with the same scaffold in a consistent manner. However, I’ve noticed a problem occurring with molecules that contain defined stereogenic double bonds (E or Z configuration).
In the example workflow attached at the end of this post the 2D representation (i.e. the rendered image) of (Z)-tamoxifen is altered to (E)-tamoxifen, even though the double bond tag remains as “(Z)” (see image below).
This discrepancy raises concerns about the node potentially changing stereochemistry during the coordinate generation process.
Is this behaviour expected when using the RDKit Generate Coords node, or might I be overlooking something in my configuration? Any insights or solutions would be greatly appreciated.
Cheers,
Gio
rdkit_generate_coords_issue.knwf (82.1 KB)