Hi Markus,

Curt's instincts are dead on: the problem here is the rings.

I'll show the fix and then explain what's going on. You just need to add
one line to your code:

core = "[a]12[a][a][a][a][a]1[a][a][a]2"
pattern = Chem.MolFromSmarts(core)

when I do this, I get the following depiction for "c1(ocn2)c2cccc1":

(The highlighting is due to the substructure match that's done during the
generation of coordinates).

So why is this necessary?
The code that generates 2D coordinates uses information about the size of
ring systems in the molecule as part of the coordinate generation. If no
ring information is present (which is true of molecules generated from
SMARTS since they are not fully sanitized on construction) then the code
calls FastFindRings(). This function is perfectly capable of identifying
all ring atoms and bonds, but it isn't very good at getting ring sizes
correct for fused systems (it finds rings, but not the smallest rings). The
consequences are the badly generated coordinates for fused ring systems
that you were seeing.

I think the current behavior of the code "isn't really ideal": the
coordinate generation code should call the SSSR algorithm in these cases so
that it can generate better coordinates. I'll take a look at the code and
think about changing it.

As an aside: if you're puzzled by the behavior of
AllChem.GenerateDepictionMatching2DStructure() you can always just take a
look at the drawing of the query molecule itself. It's not always the most
informative depiction when it comes to what the atom and bond queries are,
but you at least will see the coordinates.

A second aside: the molecule depictions in that notebook indicate that you
are stuck using the fallback drawing code, which creates fairly ugly
pictures. You can get better drawings by either installing cairo and
pycairo (in which case the code should automatically use those) or telling
the drawing code to use SVG for the rendering:

from rdkit.Chem.Draw import IPythonConsole

It really does make the drawings a lot better.

I hope this helps,

On Wed, Sep 21, 2016 at 8:47 PM, Markus Metz <metm...@gmail.com> wrote:

> Hello all:
> I am trying to perform a 2D alignment of molecules by using a pattern for
> which I am using Compute2DCoords.
> If I use a smarts string matching napthalene the 2D depiction is as one
> would expect.
> However, if I am switching to a 5,6 aromatic smarts pattern the matched
> benzoxazol the 2D structure looks rather unusual.
> Is there a way to match the 5,6 with the 6,6 pattern behavior?
> Any hint is very much appreciated,
> Markus
> P.S. a work book is attached.
> ------------------------------------------------------------
> ------------------
> _______________________________________________
> Rdkit-discuss mailing list
> Rdkit-discuss@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/rdkit-discuss
Rdkit-discuss mailing list

Reply via email to