From: Otis R. <osr...@ch...> - 2010-11-12 01:37:54
|
All- I'm looking for some advice from inorganic chemists (or others) on the subject of appropriate molfiles for coordination compounds. To put the question in perspective, we are assembling a database of organic compounds found in standard organic text. This list will, of course, include some coordination compounds. The database will include, _A_ SMILES that Jmol creates for the model, _THE_ unique SMILES that NIH Resolver creates, a common name, the IUPAC name, and a hook to the molfile. It is the SMILES & molfile that present a problem for coordination compounds. What is the standard (if any exists) for creating coordinate bonds in a molfile? It appears that a common approach is to not show bonds at all (PubChem and Emolecules seem to do this for example), with the resulting SMILES being a multi-structure SMILES. For our purpose, a single bound geometry retaining entity is preferred. We can create the unconnected model in Spartan, and then append/edit the molfile connection table to show coordinate bonds. Is there a preference or some standard for the molfiles of these compounds? Single, partial, unidentified are among the molefile options. Bob is the master of Jmol SMILES, so I'll direct the second question to him. The SMILES on these compounds present another problem in Jmol. By way of example, if we construct an unconnected ferrocene in SPARTAN and then edit the molfile connection table so that the cyclopentadienyl anion carbons are assigned "aromatic" and the coordinate bonds to iron are assigned as type 8 ("any", evidently interpreted by Jmol as partial ), then we get the following SMILES from Jmol: [C@H]12[Fe]3456789%10[C@H]%11[C@@H]%10[C@@H]9[C@@H]8[C@H]%117.[C@H]16[C@@H]5[C@@H]4[C@H]23 The NIH Resolver reads this and returns evidence that it understands the bond connections - correctly connected 2D molfile and correctly connected GIF image. This appears to be a correct SMILES for the ferrocene that we constructed. It appears that the SMILES is very confused about the geometry of the cylopentadienyl anion hydrogens, but the connections are correct. Unfortunately, Jmol does not seem to recognize what it created. With the same model in the Jmol window, the following : var x = "[C@H]12[Fe]3456789%10[C@H]%11[C@@H]%10[C@@H]9[C@@H]8[C@H]%117.[C@H]16[C@@H]5[C@@H]4[C@H]23"; var y = {*}.find("SMILES",x);echo @y does not produce a match. Resolver will also produce a unique SMILES from Jmol's SMILES: [C@H]12[C@@H]3[C@@H]4[C@@H]5[C@H]1[Fe]23456789[C@H]%10[C@@H]6[C@@H]7[C@@H]8[C@@H]9%10 The following SMILES Find (Resolver in Jmol) is a perfect match: var x = "[C@H]12[C@@H]3[C@@H]4[C@@H]5[C@H]1[Fe]23456789[C@H]%10[C@@H]6[C@@H]7[C@@H]8[C@@H]9%10"; var y = {*}.find("SMILES",x);echo @y Help! Otis -- Otis Rothenberger chemagic.com |