On Thu, Apr 27, 2017 at 9:54 AM, John Mayfield <john.wilkinson...@gmail.com>
wrote:
> I was wondering does anyone use n-way bonds in CDK (a bond with more than
> two atoms)?
>
> IAtom atom1 = object.getBuilder().newInstance(IAtom.class, "C");
>> IAtom atom2 = object.getBuilder().newInstance(IAtom.class, "O");
>> IAtom atom3 = object.getBuilder().newInstance(IAtom.class, "C");
>> IAtom atom4 = object.getBuilder().newInstance(IAtom.class, "C");
>> IAtom atom5 = object.getBuilder().newInstance(IAtom.class, "C");
>>
>> IBond bond1 = new Bond(new IAtom[]{atom1, atom2, atom3, atom4, atom5});
>>
>> I don't deny they exists but not sure how useful they are as they don't
> fit into traditional valence bond theory. Many useful algorithms break down
> (e.g. simple traversals, substructure matching, and canonical labelling).
> Rich Apodaca talks about it a little in FlexMol
> <http://depth-first.com/articles/2006/12/20/a-molecular-language-for-modern-chemistry-getting-started-with-flexmol/>
> and
> shows it for doing Pi bonding - nice but again not really practical. I seem
> to recall Egon saying CML can "store" them but I'm struggling to see if
> it's anything more than nicety.
>
It's a chicken and egg problem... people don't use it, and even can't use
it, because there are no good tools yet; not making good tools also ensures
no one uses it.
One of the reasons why we have incorrect data, is because at some point
someone took a shortcut in some tool, because it covered most of the issues.
> I believe a better way to handle (if you wanted to) would be with a higher
> level description as sets/collections of atoms/bonds over a wrong but
> acceptable valence bond graph:
>
We don't do much with transition states, but one of the key advantages of
the CDK (over RDKit, OpenBabel) is that you actually can... Fairly, I did
not track with literature citing the CDK actually uses these features...
So, besides the obvious representation (and viz) use cases below (just
think of what mess some PubChem/ChemSpider entries are), these kind of
features are essential for reaction mechanisms, so the first place to check
would be the MaCIE work and the metabolite identification cheminformatics
work from Chris' lab in the past.
Other use cases, one that I never found time for, is (also obvious from the
below) cheminformatics for organometallics... there is a good bit of
literature doing VS of such substances, typically severely limited in the
cheminformatics they use...
> Three-center one-electron valence bond graph (via CDK Depict)
> [image: Inline images 2]
> Using multicentre Sgroups (via CDK Depict)
> [image: Inline images 3]
>
So, rather than trunking the CDK, I would suggest, let's make work of
convincing the cheminformatics community of these advanced features, just
like others are doing to the PTM-protein sequence mashups... let's be a
leader, rather than a follower.
Egon
--
E.L. Willighagen
Department of Bioinformatics - BiGCaT
Maastricht University (http://www.bigcat.unimaas.nl/)
Homepage: http://egonw.github.com/
LinkedIn: http://se.linkedin.com/in/egonw
Blog: http://chem-bla-ics.blogspot.com/
PubList: http://www.citeulike.org/user/egonw/tag/papers
ORCID: 0000-0001-7542-0286
ImpactStory: https://impactstory.org/u/egonwillighagen
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Cdk-user mailing list
Cdk-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/cdk-user
