Hi,
Is there anyway to allow the atom signatures node to output the atom number as well as the signatures. Currently it is not easily obvious which signature corresponds to which atom.
Simon.
This was the same issue with the nmr workflow where one need atom numbers. This issue was also discussed in the indigo/rdkit forums. I dunno why noone seems to implement returning atomnumbers by default when substructures are matched. It's very useful
Hi,
Luis, who developed the atom signature node, will implement an option to return the atom numbers in a separate column next to the signatures by the end of this week. Complementary to that, we have added a global option to display atom ids (see this post).
Stephan
Hi many thanks for introducing an atom numbers column in the atom signatures node, this is really useful. Unfortunately it seems to have introduced a bug.
When you set signature height from 1 to 6, it returns the columns as "Signature 2" to "Signature 7". Previously these columns were "Signature 1" to "Signature 6".
Any ideas what is happening here ?
Simon.
Hi Simon,
I could reproduce the error and fixed it. The signature indices should now be displayed correctly.
The newly introduced atom id column had set the counter off by one.
Stephan
Hi Stephan,
Thanks the column renaming, it is now fixed.
I have a question about the Hydrogen atom signatures functionality.
It seems its only possible to get this to work, when the Hydrogens have been populated around the atoms, and there is no functionality in CDK to do this, I am having to use Indigo nodes (Hydrogen Adder) and then convert back. I am unsure why it is a requirement to populate the hydrogens around the molecule, I have tried the CDK Hydrogen Manipulator node first, but this doesnt help, the Signature node just reports back this error; ERROR Atom Signatures Execute failed: String value can't be null.
Also as a lot of the hydrogens on a molecule are equivalent, rather than give a unique atom number per hydrogen, would it be better to give the atom number of the hydrogen to correspond to the heavy atom it is attached too (i.e. report the Carbon number it is attached to).
I have also noticed that the Hydrogen Atom Signature node fails for divalent Sn atoms, as its expecting hydrogens when this is not necessarily the case. It reports back,
ERROR Atom Signatures Execute failed: String value can't be null.
Simon.
Hi Simon,
right you are. There was a bug introduced by the recent changes in the node.
You shouldn't have to worry about the hydrogens as this should (and now will) be handled under the hood by the node. Following your advise I have modified the node according to your description. There certainly is no point of displaying the same hydrogen signature or HOSE code three times, e.g., for a methyl group.
The build should be ready tomorrow.
Best regards,
Stephan
Hi Stephan,
THanks for the changes to the Atom Signatures node, this is working absolutely brilliantly now. Really happy with this.
Its working great for creating 1H NMR NMR predictions!!
Are there any plans to extend atom signatures to other atoms such as F, N, O, B, Si, S, P ? This would allow NMR predictions of other atom types. If you do consider implementing these other atoms, it would be useful to allow the node not to fail if it encounters molecules which doesnt have one of the atom types (i.e. it passes an empty table), as this is much more likely to occur with these atoms than C or H.
Many many thanks,
Simon.
Hi Simon,
the node dialog lets you choose either of the following elements: C, H, F, N, O, B, Si, S, P.
If no atom in a molecular structure corresponds to the selected element, the row is returned with missing values for the signature / hose code cells. Subsequently you can filter out missing values.
I am a bit reluctant to return an empty table or filter out rows that contain structures for which no signature / hose code can be generated as this would, in a way, contradict the basic idea of a workflow.
I would be grateful if you could test the node with your dataset as well and see if all works okay.
Stephan
P.S.: Rather exotic elements for your typical NMR. :)
Thanks so much for all these implementations, I look forward to trying the update out in tonight’s nightly build.
I agree with your comments that reporting missing values would be better than filtering out rows. Thanks for thinking of that option.
This has now become a very powerful and versatile node!
As the open source nmrshiftDB contains spectra for these other atoms it will be great now to build some NMR predictions on these other atom types. Just adds an extra tool to the chemists for analysing and interpreting molecules!
Simon.
Hi Stephan,
Thanks for the other atoms added to the Atom Signatures node, I am now generating exotic NMR predictions !!
Its working great!
Simon.
This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.