Chemical Transformation wiping structures

Hi, I am using the chemical transformation node to try and protonate an deprotonate my salt-stripped compounds. 

However, when quaternary nitrogens are passed through, they get wiped and the message is:

WARN      RDKitChemicalTransformationNodeModel     The following result product could not be sanitized: C(O)(CCN(C1=Nc2c(cc(Cl)cc2)C(c2ccccc2)=N(O)C1)C)=O

WARN      RDKitChemicalTransformationNodeModel     The following result product could not be sanitized: C1N(c2ccccc2)=NC(c2ccccc2)=NN1c1ccc(C)cc1

Why is it doing this?

It looks like it is converting the molfile that is being passed through into SMILES and then it's losing the charge on the nitrogens?

I am losing thousands of compounds this way so I would appreciate finding out what is happening. 

 

thanks

Louisa

Hi Louisa,

the console output that you see is a SMILES, but purely for informational purposes. The actual transformation is done on the RDKit molecule which gets created from whatever you use as input. I am only creating the SMILES from the product that could not be sanitized to have it in a short format for the KNIME console. 

Are you using an RDKit From Molecule node before in the workflow to create the RDKit molecule from an SDF mol block? If not, you may want to try it so that you are able to control the conversion in more detail. Maybe the default conversion that is done on the fly when connecting an RDKit node directly to a table with an SDF column is doing something that screws the molecules you have trouble with.

Kind regards,

Manuel Schwarze 

Hi Manuel

 

Yes, I am converting using RDKit from Molecule. it only seems to be the quaternary nitrogens. Once i remove them from my compound set (~78,000), I have no issues.

It's purely this subset of compounds....

thanks

Louisa,

This is going to be tricky to track down without a bit more info. Any chance you could attach a sample workflow with one or two input molecules that shows the problem?

-greg

Hi Greg

Here is an SDfile of some of the compounds which failed.

Out of the 20,000+ compounds that went through, about 4,000 failed this transformation.

The RXN files are just simple 'add H' and 'remove H' rxns to cover different compound types

I have also included an sdfile (notsanitised) of a selection of compounds and their smiles from our database that fail.

thanks

Louisa

Similar compounds also seem to be kicked out when using the RDKit to Molecule node.

205 compounds in my set (81,000) are kicked out with a 'MolSanitizeException'.

I have attached a picture of some of these compounds.

Any help with this is appreciated.

thanks

Louisa

I remember there being a discussing in a different context about RDKit and especially metals, resp. organometallic compounds - the solution was unfortunately referring to the commercial Chemaxon kit.

 

Is there any update for this?

thanks