The CDKAtomTypeMatcher perceives a type "Se.2", while there is no such type in the ontology
since the case statement looks fine in the atom type matcher (in the sense that a single selenium atom does occur), presumably it got lost from the great atomtype patch flood of 2011.
Patch with definition and test case
Aargh. The reaction that failed has a different selenium (H2Se):
http://www.genome.jp/dbget-bin/www_bget?rn:R03595
while the original file had just "Se". So perhaps the 'Se.2' atomtype doesn't exist after all?
OK, I searched ChemSpider for neutral Se's with just one neighbor but that does not find any...
So, shall we remove Se.2 then?
Hi, I get the same error on this compound:
http://www.hmdb.ca/metabolites/HMDB04119
with SMILES NC(CC[SeH])C(O)=O in cdk 1.5.0 from rcdklibs-1.5.0
Yours, Steffen
Increased priority - this needs a definitive decision. Personally I have found this error many times.
additional patch: https://sourceforge.net/p/cdk/patches/558/
So, just to make clear, this is the single selenium atom type, with no neighbors at all.
Do we want to have this atom type?
selenide?
Ha... well, "of course" I meant an uncharged selenium atom without any neighbors :)
Selenide is actually a different atom type, which we also do not have, according to my Groovy Cheminformatics book.. (the table in the appendix is really handy...)
Patch to add the 'selenide' atom type -> https://sourceforge.net/p/cdk/patches/560/
I update Gilleain's patch and integrated into the latest cdk-1.4.x code:
https://sourceforge.net/p/cdk/patches/563/
Has this been applied to the master (ie 1.5) branch?
Yes, this patch has been applied.