RE: [InChI-discuss] Mixtures not permitted?

[Alan W. <alan.wood@JUSTIS.COM>]

Peter

You, I and Dmitrii are all agreed that InChIs are not supposed to be used
for mixtures, apart from the special cases.

The problem is that it is very simple to generate an InChI for any mixture,
and there does not seem to be anything official that states that this is not
valid.

It is apparent that some people (e.g. PubChem) are generating InChIs for
mixtures.  Without a statement from IUPAC, how can we convince them that
they are wrong?

I have the highest regard for your FAQ, but if I cite it as evidence that
InChIs for mixtures are not valid, then someone who is happily using InChIs
for mixtures can easily point out that it is "Unofficial".

I am surprised that the Technical Manual and User Guide do not specifically
state that mixtures are not supported.

How do we get the documentation amended?

-- 
Alan Wood
http://www.alanwood.net (Unicode, special characters, pesticide names)
 

> -----Original Message-----
> From: Dr P. Murray-Rust [mailto:pm...@ca...]
> Sent: 21 March 2006 11:24
> To: inc...@li...
> Subject: Re: [InChI-discuss] Mixtures not permitted?
> 
> 
> On Mar 21 2006, Alan Wood wrote:
> 
> >I am sure that I have read somewhere that the current 
> version of InChi is
> >not to be used for mixtures (except racemic mixtures), but 
> the only place I
> >can find it stated is in Nick Day's Unofficial FAQ.
> >
> >As far as I can see, the documentation supplied with the 
> InChI generation
> >program does not say that mixtures are not supported.
> >
> >Please can someone tell me where it is officially stated 
> that you cannot
> >have an InChi for a mixture?
> >
> >InChIs for mixtures are easy to generate, and they are 
> appearing on the
> >Internet, for example:
> >
> ><http://pubchem.ncbi.nlm.nih.gov/summary/summary.cgi?cid=62261>
> ><http://pubchem.ncbi.nlm.nih.gov/summary/summary.cgi?cid=162382>
> >
> As you say the FAQ is unofficial, so anything Nick and I say 
> is in that 
> light...
> 
> It depends what you mean by mixture. InChI does support the integer 
> compostion of components - e.g. hydrates, stoichiometric 
> clathrates, etc. 
> What IMO it does not support is arbitrary fractional combinations of 
> components. The only logical interpretation possible of the 
> examples above 
> is that there are equimolar amounts of the components. I 
> doubt that is the 
> case - I suspect that these entries represent a mixture of 
> components in 
> unknown and probably variable amounts. If so I think it is a 
> dangerous 
> extension of InChI which should map to a precise species 
> representatable by 
> one or more connection tables.
> 
> It is impossible to canonicalise variable non-integral 
> components which is 
> why InChI should not, for example, try to carry other 
> information such as 
> cell parameters or other measurements. InChI is an identifier 
> based on 
> precisely known composition and (usually) precisely know 
> connection tables. 
> Any attempts to extend it - or to make it carry additional 
> information - 
> without very thorough _public_ analysis are likely to break. 
> We don't even 
> have universal software yet for comparing InChIs at different 
> levels of 
> representation.
> 
> But that is all unofficial.
> 
> P.
> 
> 
> -- 
> Peter Murray-Rust
> Unilever Centre for Molecular Informatics
> Chemistry Department, Cambridge University
> Lensfield Road, CAMBRIDGE, CB2 1EW, UK
> Tel: +44-1223-763069 Fax: +44 1223 763076 
> 
> 
> 
> -------------------------------------------------------