The reply that Ling forwards has one approach to doing this.
It's a bit easier for someone who is willing to do some C++ work.[1]
One could imagine writing a function prepareForResonanceFormMatching(ROMol
&m) (or some such thing) that would be applied to the *query* molecule that
does the following:
- identifies the groups that need to be resonance-symmetrized
- changes the resonance bonds to Query bonds that match single or double
(possibly also aromatic?)
- neutralizes any charges on resonating atoms in the group.
The last step is important because the query C(O)O matches the molecule
C(O)[O-] twice, but C(O)[O-] only matches once:
In [11]:
Chem.MolFromSmiles('C(O)[O-]').GetSubstructMatches(Chem.MolFromSmiles('C(O)O'),uniquify=False)
Out[11]: ((0, 1, 2), (0, 2, 1))
In [12]:
Chem.MolFromSmiles('C(O)[O-]').GetSubstructMatches(Chem.MolFromSmiles('C([O-])O'),uniquify=False)
Out[12]: ((0, 2, 1),)
I suspect such a function would be useful to multiple people.
For identifying the groups that are resonance symmetrized: though this
could be done using a set of particular patterns, it may be better to think
about doing it more generally by having it find resonance systems.[2] The
flag Bond.getIsConjugated(), set during sanitization, is probably useful
for this.
-greg
[1] well, to the extent that anything is ever "easier" in C++
[2] this would allow finding the substructure matches within molecules
like C1=C(C)C=CC=CC=C1
On Fri, Oct 31, 2014 at 2:09 AM, S.L. Chan <slch...@yahoo.com> wrote:
> Dear Paolo,
>
> I have asked a very similar question last year. This was what Greg said.
>
> Ling
>
> Re: [Rdkit-discuss] atom equivalence for substructure matching
> <http://www.mail-archive.com/rdkit-discuss%40lists.sourceforge.net/msg03515.html>
>
>
> [image: image]
> <http://www.mail-archive.com/rdkit-discuss%40lists.sourceforge.net/msg03515.html>
>
>
>
>
>
> Re: [Rdkit-discuss] atom equivalence for substructure ma...
> <http://www.mail-archive.com/rdkit-discuss%40lists.sourceforge.net/msg03515.html>
> Skip to site navigation (Press enter)
> View on www.mail-archive.com
> <http://www.mail-archive.com/rdkit-discuss%40lists.sourceforge.net/msg03515.html>
> Preview by Yahoo
>
>
> ------------------------------
> *From:* Paolo Tosco <paolo.to...@unito.it>
> *To:* "rdkit-discuss@lists.sourceforge.net" <
> rdkit-discuss@lists.sourceforge.net>
> *Sent:* Thursday, October 30, 2014 4:26 PM
> *Subject:* [Rdkit-discuss] GetSubstructMatches() and resonance structures
>
> Dear all,
>
> The following code snippet compares two resonance structures of formate
> anion:
>
> import rdkit
> from rdkit import Chem
>
> mol1=Chem.MolFromSmiles('C([O-])=O')
> mol2=Chem.MolFromSmiles('C(=O)[O-]')
> mol1.GetSubstructMatches(mol2, uniquify = False)
> ((0, 2, 1),)
>
> mol1.GetSubstructMatches(mol1, uniquify = False)
> ((0, 1, 2),)
>
> I would rather like to get, in both cases, the following output:
> ((0, 1, 2),(0, 2, 1))
>
> which would account for the carboxylate group symmetry due to resonance.
> The same applies to amidinium, guanidinium, etc.
>
> Is that currently feasible within the RDKit API?
>
> Thanks in advance, cheers
> Paolo
>
>
>
> ------------------------------------------------------------------------------
> _______________________________________________
> Rdkit-discuss mailing list
> Rdkit-discuss@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
>
>
>
>
> ------------------------------------------------------------------------------
>
> _______________________________________________
> Rdkit-discuss mailing list
> Rdkit-discuss@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
>
>
------------------------------------------------------------------------------
_______________________________________________
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
