On Wed, Jun 23, 2010 at 7:43 PM, Craig A. James <cja...@emolecules.com> wrote: > On 6/23/10 9:44 AM, Tim Vandermeersch wrote: >>> Sorry to keep nagging, but the remaining test failure seems to be real >>> bug: http://my.cdash.org/testDetails.php?test=2416695&build=76838 >>> >>> It says that babel -ismi -osmi >>> '[nH]1c2ccccc2c2c3C(=O)NCc3c3c4ccccc4[nH]c3c12' gives >>> '[nH]1c2CCCCc2c2c3C(=O)NCc3c3c4ccccc4[nH]c3c12'. >>> >>> If you look at both in Daylight Depict, you will see that aromaticity >>> is lost in a phenyl ring at the end of five conjugated rings. >> >> The SSSR and LSSR are the same (6 rings) for this molecule. The >> problem must be in Craig's new aromaticity code... > > As far as I know, I didn't change the aromaticity algorithm, only the > efficiency of applying it. The actual rules should be identical (and still > broken as shown by this example).
The failing test case is a simple benzene ring (attached to 4 other aromatic rings). Electron counting isn't an issue here AFAIK. > The solution requires a complete rewrite of the rules. We discussed this > recently in the OpenSMILES discussion. The problem (IMHO) is the poor > definitions of electron counts for each SMARTS. Some SMARTS are allowed zero > or one electron, some are allowed one or two electrons. That's completely > bogus if you ask me. It just means your SMARTS aren't specific enough. Right, the actual aromaticity model doesn't matter at all. The only requirement for writing canonical smiles is that aromaticity can be assigned canonically. It doesn't have to make chemical sense... > We need to nail down a clear, concise ALGORITHMIC definition of aromaticity, > including the SMARTS that define the electron contributions of every possible > aromatic atom and its environment. I've been nagging everyone for about two > years about this, with little progress. Yes. I don't know where the current rules come from and there is little to be found on this topic in literature. However, a great step forward would be to compose a test set of structures we wish our algorithm to classify (Noel suggested this). Ideally most cases make chemical sense but unambiguous aromaticity assignment has a higher priority. Even if some cases are "obviously wrong" to chemists. We might want to include other projects/people if this will become the OpenSMILES aromaticity model. > We should stop writing code and finish writing the specification. Until we do > that, we'll keep patching specific cases and arguing about it. Some code to verify the specification is probably needed but I agree a specification must be chosen here. Tim > Craig > > ------------------------------------------------------------------------------ > ThinkGeek and WIRED's GeekDad team up for the Ultimate > GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the > lucky parental unit. See the prize list and enter to win: > http://p.sf.net/sfu/thinkgeek-promo > _______________________________________________ > OpenBabel-Devel mailing list > OpenBabel-Devel@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/openbabel-devel > ------------------------------------------------------------------------------ ThinkGeek and WIRED's GeekDad team up for the Ultimate GeekDad Father's Day Giveaway. ONE MASSIVE PRIZE to the lucky parental unit. See the prize list and enter to win: http://p.sf.net/sfu/thinkgeek-promo _______________________________________________ OpenBabel-Devel mailing list OpenBabel-Devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/openbabel-devel
