On Thu, Feb 19, 2015 at 6:54 PM, Dimitri Maziuk <dmaz...@bmrb.wisc.edu>
wrote:
> On 02/19/2015 08:54 AM, Markus Sitzmann wrote:
> > A database can have several definitions of unique for anything - a
> > structure database can have this, too. If you have a chemical compound
> > which can form 10 different tautomers, you can represent the compound
> > by 10 chemical structures (it is still the same compound, though). So,
> > if you define uniqueness on basis of chemical compound, you have one
> > db entry and this one entry has a single (tatuomer-sensitive) InChI
> > covering 10 chemical structures; if you define uniqueness on basis of
> > tautomers/chemical structures (because all are relevant, for instance,
> > in NMR spectrosopy) you have (and want) 10 database entries, each with
> > a single (tautomer-sensitive) InChI. Two definitions of unique.
>
> No. there's only one definition if unique: unique key is a set of
> attributes that is guaranteed to be unique for each entity. The
> relationship between the key and the entity is symmetric: if x is the
> inchi string for compound y then y is the compound for inchi string x.
>
This could be true if the InChI algorithm just took the input structure you
provided and returned a string for it. That's not, however, what it does.
The algorithm first standardizes the molecule: modifying the tautomeric
state, moving charges around, etc., then generates the InChI string.
Assuming you use the full standard InChI string, what you end up with is a
function that maps every molecule to a unique InChI. You can also map every
InChI back to a unique molecule (though, as discussed above and elsewhere,
this can be somewhat fraught with danger), but the unique molecule that the
InChI corresponds to is not necessarily the same as the input molecule.
-greg
------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk
_______________________________________________
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
