Unusual cd_smiles entries in cd_smiles

Hello,

after cartridge upgrade from 5.10.2 to 14.7.14 there are entries in cd_smiles like this:

[#6]\[#6](-[#6])=[#7]\[#8]-[#6](=O)-[#7]-c1ccccc1

while on the older version it was CC(C)=NOC(=O)Nc1ccccc1

Isn't the first one is a smarts not a smiles? Why was it changed like this?

I also have several records, where cd_smiles is null, even though on the older version there wasn't any.

Thanks,

Istvan

Dear István,

 

The behaviour described could happen easily; cd_smiles/cd_smarts columns are meant to be used internally by Chemaxon products. They store the standardized form of the original structure in one of the formats: (cx)SMILES, (cx)SMARTS. It has NULL values when the structure could not be represented in none of the formats listed above. We have to admit that the naming is not the best, the cd_standardized would be more informative. 

The reason for the NULL values in your newer table is most probably because of the more sophisticated checking mechanisms we employ in the newer version compared to the 5.10.2 one.

 

All the changes listed above shall not affect your search behaviour. If you experience differences, please let us know.

 

Best regards,

Norbert.

Dear István,

 

Could you please send us a structure in its original format (content of the cd_structure field) that is affected by this change. We would like to examine whether transforming it into SMARTS or NULL is correct.

 

Thank you!

 

Best regards,

Norbert

Hello,

I attach three (named invalid1, 2 and 3) mol files in the zip, all of them has null value in cd_smiles.

The other zip contains a mol file, that has a smarts in cd_smiles (look in the cxsmi file).

Best regard,

Istvan

Dear István,

 

Thank you for the structures sent. We have succeed to reproduce the behaviour described, the structures are converted into SARTS or NULL even when they have valid SMILES representation. We will fix this problem. Fortunately - based on our tests - this does not affects the correctness of the search.

 

We will let you know when the issue is solved.

 

Best regards,

Norbert

Dear István,

 

We have investigated the structures enclosed in invalids.zip and concluded that one (out of 3) structure could be represented in SMILES without loosing any structural information, while the rest would keep all the structural information (mostly stereo) only in SMARTS format. As a consequence:

- for 2 out of 3 structures the cd_smiles field will contain SMARTS, 

- for the third one the cd_smiles field will contain SMILES 

- there should be no NULL value present for these structures.

 

The changes are already available from version 14.9.22.0

 

Best regards,

Norbert