Re: [Rdkit-discuss] Count carbon atoms

[Greg L. <gre...@gm...>]

On Wed, Oct 7, 2015 at 11:12 AM, Joos Kiener <joo...@gm...> wrote:

> Hi all,
>
> is there an easy way I'm missing to get the number of C-Atoms in a
> molecule?
>
> Currently I iterate all atoms and check if it's symbol is C. Doesn't seem
> very efficient.
>

There's an, obviously under-documented, simple (and fast) way to do this:

from rdkit.Chem import rdqueries
q = rdqueries.AtomNumEqualsQueryAtom(6)
len(mol.GetAtomsMatchingQuery(q))

This is, based on the testing I just did, significantly faster than either
the substructure-based or GetAtoms() based approaches. I'm not sure how
generally useful it is, but this could be made even faster by adding a
Mol.CountAtomsMatchingQuery() function to the C++ interface.

You can find a bit of explanation, as well as a list of the types of
QueryAtom functions available, by searching for QueryAtom here:
https://github.com/rdkit/UGM_2014/blob/master/Notebooks/Whats_new.ipynb

Best,
-greg