Menu
▾
▴
#1308 cannot parse SMILES due to "SMILES strings cannot ring close two directly connected atoms"
beta-CYCLODEXTRIN (http://pubchem.ncbi.nlm.nih.gov/summary/summary.cgi?sid=160657156&viewopt=PubChem) could be parsed with 1.4.7, but fails with 1.4.14
new SmilesParser(DefaultChemObjectBuilder.getInstance()).parseSmiles("OC1O2O3(CO)O(O4(CO)O(O5C(CO)O(O6(CO)O(O7(CO)O(O8(CO)O(O1(O)2O)(O)8O)(O)7O)C(O)6O)(O)5O)(O)4O)(O)3O");
correction: dont know about 1.4.14, but it fails with 1.4.18
Hi Martin,
Daylight won't read it either.
http://www.daylight.com/daycgi/depict?4f43314f324f3328434f294f284f3428434f294f284f354328434f294f284f3628434f294f284f3728434f294f284f3828434f294f284f31284f29324f29284f29384f29284f29374f2943284f29364f29284f29354f29284f29344f29284f29334f
I'm guessing it was computed with CDK?
Using the compound id smiles works okay: C(C1C2C(C(C(O1)OC3C(OC(C(C3O)O)OC4C(OC(C(C4O)O)OC5C(OC(C(C5O)O)OC6C(OC(C
(C6O)O)OC7C(OC(C(C7O)O)OC8C(OC(O2)C(C8O)O)CO)CO)CO)CO)CO)CO)O)O)O
from http://pubchem.ncbi.nlm.nih.gov/summary/summary.cgi?cid=444041
Also, don't use DefaultChemObjectBuilder, SilentChemObjectBuilder is better (yes it should be other way round). The default one fires events on every update… which you don't need unless it's a GUI app. This can have massive hits in performance for some algorithms.
J
On 28 Jun 2013, at 14:03, "Martin Gütlein" martinguetlein@users.sf.net wrote:
Related
Bugs:
#1308Hmm, I don't actually know, I have it from a project partners database..
Okay, I will switch to the Silent one. It does not read this specify compound as well, though
OpenBabel accepts it, BTW
Okay here was the commit that changed it: https://github.com/cdk/cdk/commit/29d6fbbe963393b3fd4d6a4caf5e5a8ae8b818e4
The SMILES is not valid, here is what OpenBabel reads:
http://apps.ideaconsult.net:8080/ambit2/depict?search=OC1O2O3%28CO%29O%28O4%28CO%29O%28O5C%28CO%29O%28O6%28CO%29O%28O7%28CO%29O%28O8%28CO%29O%28O1%28O%292O%29%28O%298O%29%28O%297O%29C%28O%296O%29%28O%295O%29%28O%294O%29%28O%293O&smarts=
The CDK version there hasn't been updated so still reads, but you can see it's not the correct molecule. Basically the smiles try to make a 2 member ring, which is impossible - the newer code checks for these cases and throws an exception.
On 28 Jun 2013, at 14:35, "Martin Gütlein" martinguetlein@users.sf.net wrote:
Related
Bugs:
#1308okay, thanks, I think we have to fix the original smiles, then
I dont know how to close this ticket, could you do that for me?
Yeah, sourceforge change it so the submitter couldn't do that anymore - don't know why.