« One Conspiracy After Another |
| Protein-Protein Compounds - The Flip Side »
June 25, 2013
Picking Out Incorrect Natural Products
Once in a while, you see people who've gone to the trouble of synthesizing a natural product, only to find that its structure had been incorrectly assigned. (Back in the days when structure elucidation was much harder, R. B. Woodward had this on his list of reasons to do total synthesis, although it wasn't number one).
Now there might be computational method that could flag incorrect structures earlier. This paper describes a carbon-13-NMR-based neural-network program, from a training set of 200 natural products, that seems to do a good job of flagging inconsistencies. It won't tell you that the assigned structure is right (there's probably a list of plausible fits for any given NMR), but it will speak up when something appears to be wrong.
And that's the mode I see this being used in, actually. I suspect that some groups will be motivated to go after the misassigned compounds synthetically, if they can come up with a believable alternative, in order to revise the structure. I'm not sure what happens if you put one of those South Pacific marine toxins into it, the ones that practically need a centerfold to publish their structures in a journal, but this looks like it could be a useful tool.
+ TrackBacks (0) | Category: Analytical Chemistry | Natural Products
POST A COMMENT
- RELATED ENTRIES
- Merck's Aftermath
- Models and Reality
- Rewriting History at the Smithsonian?
- The FDA: Too Loose, Or Appropriately Brave?
- More Magic Methyls, Please
- Totaling Up a Job Search
- Humble Enzyme Dodges Spotlight
- Unraveling An Off-Rate