Sometimes two dimensions just aren't enough



  • I was browsing a chemical catalog and found this curious substance. The description says that the first item is a buckyball*, but the image says something else entirely. I guess the structure doesn't flatten very well. The second item, however, is perfectly happy to be rendered in two dimensions.



    *A buckyball should be a spherical molecule composed of 60 carbon atoms.



  • Someone's probably mapping all the connections between those atoms as a graph and storing it in a database. I imagine how the schema for the tables would be.

    Wouldn't it be much easier if they just stored images or 3d models instead of trying to rebuild the molecule everytime (unless they're already storing images and that's just some unlucky sample)? I figure some advanced applications for chemical research or teaching would make a good use of such a thing, but not a catalogue.



  • The structures are probably being stored as <a href='http://en.wikipedia.org/wiki/SMILES">SMILES (ASCII) strings, which store only the connectivity between all of the atoms in a tree. Their positions in space are inferred from the connectivity. My guess is that as the system traversed the tree it had a harder and harder time placing the atoms from the 3D molecule into the 2D plane, much like how flattening a globe onto a map distorts the land masses.

    A more advanced system might store the XYZ coordinates of the atoms along with their connectivity or use some kind of energy minimization of the structures. However, SMILES has become the industry standard and these sort of problems probably don't come up often enough to justify the extra effort.



  • @DKNewsham said:

    The structures are probably being stored as SMILES

    Quite possibly; still, the devs should be aware that their system will render some atoms badly and just not render the image in a case like this. Or better yet a pre-rendered image. I mean, this is just hideous, and it shouldn't be.



  • The second one could easily evolve into dilbert's face!



  • @ounos said:

    The second one could easily evolve into dilbert's face!

    I agree, it does kinda look like dilbert's head



  • As someone who has to work with automatic chemical diagram generation I have some sympathy for the catalogue.  You are right that Fullerenes are spheroids and therefore do not flatten into diagrams - good diagrams of them tend to resemble an isometric view of a see-through football.  They can also have different numbers of atoms than 60 - 50 and 70 being common and giving slighly squashed spheroids.  There are also substituted Fullerenes which are the same just with bits stuck on.  So the writers of such software could go to great lengths to have special cases for some of these, but there are always going to a be a few that are rubbish.  In a catalogue with probably tens of thousands of compounds it is just not worth the effort for the odd special case, and anyhow it will give the odd chemist a chuckle :)



  • @MET said:

    and anyhow it will give the odd chemist a chuckle :)

    Good enough reason for me. My days of organic chemistry were largely devoid of laughter...



  • Bah there is a great search tool out there that can take the export of that as a text file and give you contextual flat-file search.


Log in to reply