On Feb 13, 2009, at 6:20 PM, George Oakman wrote:
One of the first example I have been playing with is the canonical
SMILES for Aspirin.
..
This gave me the following result:
CC(Oc1ccccc1C(O)=O)=O
But I was expecting
CC(=O)Oc1ccccc1C(=O)O)
The canonical SMILES is canonical only on the context of an
algorithm. The Daylight algorithm is different than the RDKit one is
different from the OpenBabel one is different ... . In fact, the
Daylight algorithm has changed over time to fix various problems.
When that happens, the molecules need to be re-canonicalized.
Even if you go back to the original Weininger paper, there are
ambiguities in the description which make the result implementation-
specific.
Is the book you're using "Molecular Design" by Gisbert Schneider and
Karl-Heinz Baringhaus? That came up when I searched for "canonical
SMILES" and I see it has example of aspirin with your expected SMILES.
Andrew
[email protected]
