Hi Greg,
Thanks, that makes sense. I see the table has only charged/protonated atom
types for nitrogen. Obvious once you look at it... ;)
Cheers,
Bruce
On 16 May 2017 at 06:53, Greg Landrum <greg.land...@gmail.com> wrote:
> Hi Bruce,
>
>
>
> On Mon, May 15, 2017 at 3:46 PM, Bruce Milne <bfmi...@gmail.com> wrote:
>
>> Hi,
>>
>> I've noticed that if I define a molecule as neutral, positive and
>> negative (depending on protonation) the EState indices calculated by RDKit
>> reflect the changes as expected:
>>
>> m = Chem.MolFromSmiles('CCO')
>> m_neg = Chem.MolFromSmiles('CC[O-]')
>> m_pos = Chem.MolFromSmiles('CC[OH2+]')
>>
>>
>> EState indices:
>>
>> m : [ 1.68055556 0.25 7.56944444]
>> m_neg : [ 1.56944444 0. 8.93055556]
>> m_pos : [ 1.79166667 0.5 6.20833333]
>>
>> However, when I calculate the EState fingerprints there seems to be some
>> problem with the (de)protonated oxygen as its value (and even the presence
>> of its atoms type) fails to show up. If EState.EStateIndices() can
>> calculate the value for oxygen in all three states then
>> Fingerprinter.FingerprintMol() should also be able to handle these
>> notations?
>>
>
> The EState fingerprinter uses a set of atom types defined in Table 1 of
> this publication:
> http://pubs.acs.org/doi/abs/10.1021/ci00028a014?journalCode=jcics1
>
> Atoms that don't match any of those types don't contribute to the
> fingerprint. There are no charged O atoms in the table, so those atoms
> don't get typed.
>
> The current implementation is quite literal and just uses the atom types
> that are explicitly defined in that table. It might be worth exploring a
> different scheme for defining types that handle all (or at least most)
> atoms, but that would result in a different fingerprint.
>
> -greg
>
>
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Rdkit-discuss mailing list
Rdkit-discuss@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
