I am trying to use RDKit Molecule Substructure Filter, to find structures that contain multiple specified substructures.
Here is the workflow: RDKit_substructre_fail.knwf (186.0 KB)
The substructures I am looking for are:
C1=N*=**=*1
C1=CC=CC=C1
N1C=*C=N1
N[H]
And some of the structures that are in the list (that I believe should be matches):
Fc1cnccc1C(=O)Nc1ccnn1Cc1ccccc1C(F)(F)F
CC(C)Nc1ncc(cn1)C(=O)Nc1c(C)cnn1Cc1ccc(Cl)cc1
O=C(CN1CCOc2ccccc2C1)Nc1ccnn1Cc1cccnc1
Cc1cnn(Cc2ccccc2Cl)c1NC(=O)c1cnc(nc1)C1CC1
CC(C)n1ncnc1NC(=O)c1cnc(Nc2ccccc2)nc1
OC=O.O=C(CN1CCc2ccccc2C1)Nc1n[nH]cc1-c1ccccn1
RDKit Molecule Substructure Filter configuration:
I do believe at least some (if not all) of these structures should be a match to all of the substructures.
Can someone explain what am I doing wrong since no matches are returned?
P.s. I tried aromatizing and dearomatizing both the query molecules and the molecule list, it did not change anything.
P.p.s. I am using KNIME version 4.5.0 and RDKit Nodes Feature 4.3.0.v202106041552
You have configured the RDKit Molecule Substructure Filter node to require that all the patterns match.
When I test them individually, C1=CC=CC=C1, N1C=*C=N1 and C1=N*=**=*1 give no matches, and N[H] gives 99 matches. And so overall I get no matches at the end of the workflow.
Switching to the aromaticized versions n1c*cn1,c1n****1, and c1ccccc1 respectively, seems to work for me. I get 100 matches when I test them individually, and 99 matches when I combined them all and add N[H].
Thank you for your response. I tested the option where I use aromatized versions of the smiles and it still did not work, but only because the node “Molecule type cast” fails to aromatize rings with undefined atoms. So instead I drew each aromatized from with Marvin Sketch and used the mol files of these structures to try to do the substructure search and it successfully found 100 out of 100 substructures.
