Issues with moe wash and stereoenumerator nodes

KNIME Analytics Platform
1

Hi,
There doesn’t seem an moe forum, so posted to general for now.
I have recently been trying to use the moe nodes called wash to add chiral wedges or use the stereoenumerator node. However the output structures are only showing flat bonds. For instance, the stereoenumerator node will generate 4 entries for a molecule with 2 chiral centres but all resultant molecules have flat bonds.
Also starting with molecules with wedge bonds results in the wedges being stripped.
Either I am doing something wrong or a major bug has been introduced with the latest build.
I have chosen input structures as sdf or mol format.

Simon.

2

Hi Simon,

I think the problem is in the SD format. In this format you have two options to define chirality. One is in the atom block and the second is in the bond block. This leads to the problem that sometimes the constraints are contradicting each other. To avoid this problem MOE uses the atom block exclusively to set the chiral constraints. If the renderer you use is looking at the bond block it will not show you the wedges, but the constraints are set. You can check this using the QSAR Descriptor node. If you calculate the descriptor with code "chiral" it will give you the number of chiral centers and code "chiral_u" will give you the unconstraint centers.

If these numbers indicate that the constraints have not been set or our renderer doesn't show the molecules correctly, please let me know and I'll fix it right away.

BTW: You can contact CCG support at support@chemcomp.com any time you have a question, problem or idea for improvement about the moe nodes. :-)

Best regards

Guido

3

Hi Simon,

It's me again. I checked the code and I trapped myself because my SD renderer is looking at the bond block. :-))))

I fix this for the next update. I have to rewrite this part of the Java class anyway to support 3D molecule viewing.

Thanks for pointing this out.

Best regards

Guido

4

Thanks Guido,

I tried the chiral and chiral_u and it does seem like the chirality has been set right. However I dont seem to be able to get other nodes (i.e. Indigo and RDKit) to recognise these chiralities. Is it possible to have the MOE nodes add chirality both the bond block and atom block?

Simon.

5

Hi Simon,

I will discuss this with our developers. In KNIME I'm only using the SD export function of MOE.

I have changed the renderer so that a chiral constarint on an atom is now displayed as red or green star. The reason for not using wedges is again to avoid ambiguity. Maybe I can also do something on the Java side of our extensions. I'll let you know.

Best regards

Guido

6

Hi Simon,

I am helping a colleague with something similar, and found that if you use the MOE Molecule to SMILES node on the output of the Stereo Enumeration, then the Smiles produced have the standard chirality flags (@, @@). I would expect these would then be accepted by other tools.

Agree that if you try and pass the SDF from Stereo Enumeration to other tools it doesn't retain chirality, and would be better if this could be changed.

Dave

7

Hi Dave, the post is actually on behalf of said colleague !!

Simon.