Further investigation shows that this issue is not related to the reaction code at all, it's a general SMILES canonicalisation bug I'm afraid. Consider the following:
> mol = Chem.MolFromSmiles("C1C[C@@H](CC)CC[C@@H](CC)1")
> print Chem.MolToSmiles(mol, isomericSmiles=True)
CC[C@@H]1CC[C@@H](CC)CC1
The output should be the same as the input but plugging those strings into
the daylight website's depiction tool gives chirally different molecules.
This behaviour is observed in RDKit 2013.09 with no custom patches.
Yours,
Toby Wright
--
InhibOx Ltd
On 28 March 2014 15:47, Toby Wright <toby.wri...@inhibox.com> wrote:
> Oops, forgot to mention: This is with the solution to github issue
> #233<https://github.com/rdkit/rdkit/issues/233>patched into my RDKit build.
>
> Yours,
>
> Toby Wright
>
> --
> InhibOx Ltd
>
>
> On 28 March 2014 15:43, Toby Wright <toby.wri...@inhibox.com> wrote:
>
>> Hi,
>>
>> I believe I've found a bug in the new code that deals with reactions that
>> have chirality specified for untagged product atoms. Consider the following:
>>
>> > rxn = AllChem.ReactionFromSmarts("[C:1].[C:2]>>C1C[C@@H](C[C:1])CC[C@
>> @H](C[C:2])1")
>> > m1 = Chem.MolFromSmiles('FC')
>> > m2 = Chem.MolFromSmiles('BrC')
>> > p = rxn.RunReactants((m1,m2))[0][0]
>> > Chem.SanitizeMol(p)
>> rdkit.Chem.rdmolops.SanitizeFlags.SANITIZE_NONE
>> > Chem.MolToSmiles(p, isomericSmiles=True)
>> 'FCC[C@@H]1CC[C@@H](CCBr)CC1'
>>
>> The output looks about right, the [C@@H]s are both still [C@@H] but
>> whereas before they were both being approached from around the ring now the
>> canonicalisation has us approaching one from outside the ring. Both
>> extensions from the ring should be towards and if I convert the product
>> part of the above reaction to a png I get:
>> [image: Inline images 1]
>> but in the output one is towards and the other is away:
>> [image: Inline images 2]
>>
>> Note that I can work around this, if I specify my reaction as
>> [C:1].[C:2]>>[C:1]C[C@H]1CC[C@@H](C[C:2])CC1 thus apeing the atom
>> ordering of the product RDKit will give me I get the chirality I want, at
>> least in my test cases so far.
>>
>> Yours,
>>
>> Toby Wright
>>
>> --
>> InhibOx Ltd
>>
>
>
<<inline: MadeProduct.png>>
<<inline: ReactionProduct.png>>
------------------------------------------------------------------------------
_______________________________________________ Rdkit-discuss mailing list Rdkit-discuss@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
