Tautomers fail under jc_compare using t:ff option, tdf:y idx

Hi,

We tested a large chemical catalog with over a milliion structures, and found a few (145) cases where SMILES could be matched to other smiles in the database using jcf.standardize(<smiles>, 'config:tautomerize outFormat:smiles:u'), but the index-backed jc_compare failed to match the smiles using 't:ff tautomerSearch:y charge:i radical:i stereoSearchType:i'

The attached file lists the 145 smiles as they existed in the catalog and the database.  The columns in the file are as follows:

fwd = the result ofjcf.compare(smi_1, smi_2, 't:ff tautomerSearch:y charge:i radical:i stereoSearchType:i') 

rev = result ofjcf.compare(smi_2, smi_1, 't:ff tautomerSearch:y charge:i radical:i stereoSearchType:i') 

smi_1 = first smiles

smi_2 = second smiles

tautomeric form = This is the SAME for both smiles, or  jcf.standardize(smi_1, 'config:tautomerize outFormat:smiles:u') =  jcf.standardize(smi_2, 'config:tautomerize outFormat:smiles:u')

 

Any suggestions about why the t:ff search failed when both smiles resolve to the same tautomer would be appreciated.

Thank you for sending us your observations. We will investigate your data soon.

Best regards,

Krisztina

We are now working on our new version of JChem (5.12) to be released in the next month, containing improvements of tautomer search. When running the same queries with this version of JChem currently 112 out of the sent 145 compound pairs are hits of each other.

See attached the results of searches (t:ff tautomerSearch:y charge:i radical:i stereoSearchType:i)

query: smi_1.smi target: smi_2.smi result: result_12.smi

query: smi_2.smi target: smi_1.smi result: result_21.smi

We keep on investigating the rest 33 compound pairs.

Thank you for your patience.

Krisztina

We have investigated the case of those 33 compounds that were not hits in JChem (5.12). All the 33 pairs are resonance structures (see image attached), which are not meant to be hits with the given settings. To find them you may want to use other standardization procedures (we can help you in this issue).

Best regards,

Norbert