jc_standardize does not validate output

Hi,

I am looking for a way to validate SMILES and SMARTS in JChem Cartridge.  

In the event of invalid SMILES, jc_standardize does not perform as expected.  Please see example of using the jc_standardize operator below:

SQL> SELECT jc_standardize('BOB THE BUILDER', 'sep=~ config:removeexplicitH....a

romatize~outFormat:smiles:0') AS example

  2    FROM DUAL;

EXAMPLE
--------------------------------------------------------------------------------
BOB

SQL>

SQL> SELECT jc_standardize('bob the builder', 'sep=~ config:removeexplicitH....a
romatize~outFormat:smiles:0') AS example
2 FROM DUAL;

EXAMPLE
--------------------------------------------------------------------------------

[#5&aH2]:o:[#5&aH2]

SQL>

Hi Mike,

I am sorry to say, but both of these strings do correspond to the smiles syntax.

The character string before the first space is interpreted as smiles or smarts, and the rest is taken as molecule name.

What kind of validation would you exactly have in mind?

Probably our Structure Checker product can help with some criteria, but it won't help with ultimately ambiguous cases.

Szabolcs

This function does do well at failing on bad input for every other case I have tested.

I know it's a really unique case, but I would point out that I asked the function to output SMILES, and it gave me SMARTS, so in this case, the function failed to respond properly.

I'd prefer that the function fail when it sees invalid smiles, instead of parsing a fraction of the SMILES.

~mike

Mike,

In general after the first whitespace any addition string is treaed as a data field, so can contain anything.

http://www.chemaxon.com/marvin/help/formats/smiles-doc.html#smiles_with_info

When  you mentioned "BOB THE BUILDER" on the phone I certainly did not realize this, and that "BOB" is actually a valid SMILES - some things just work better in writing :)

However "bob" is really not a valid SMILES, as the lower case denotes aromatic atoms which are not legal in this context, so we interpret it as SMARTS on the input.

We should not export it as "smiles:0" indeed.

This is a bug in 5.10.x in 5.11.x it has been fixed.

5.10:

dorant@T500 /cygdrive/c/jchem5103/bin


$ ./molconvert smiles:0 -s "bob"


[#5&aH2]:o:[#5&aH2]

5.11:

dorant@T500 /cygdrive/c/jchem5113/bin


$ ./molconvert smiles -s "bob"


 cannot convert molecule 1 to smiles: The following atom cannot be aromatic according to the SMILES definition: B

Best regards,

Szilard

Post for testing issue with notifications - please ignore.