indigo-depict: atom order, chiral SMILES

[Marcin Borkowski]

Hi,

1. Is there a way to control the way in which the molecules are
drawn by depict? What I mean is sometimes it would be more convenient
to me if the molecule was drawn exactly as entered, but depict
reverses it - for example CCCCS is drawn as SCCCC. Is this effect of
the canonicalisation? If so, is there a way to switch it off?

2. Perhaps I am not understanding SMILES or chirality, while I started
as a chemist 30 years ago today I am more of a computer guy. I have
problems understanding image created from the C[C@](O)(N)(Cl) code.
What I get is NH2 pointing up, Cl pointing left up, OH pointing left
down, and solid wedged CH3 pointing right down. Your page links to
IUPAC documents, and I hoped they will help me understand the image,
but after I have downloaded the GRAPHICAL REPRESENTATION OF
STEREOCHEMICAL CONFIGURATION I found the type of structure
indigo-depict creates is listed as not acceptable (ST-1.1.4).

I am using Indigo 1.1, downloaded and compiled just a few days ago. If
needed, I can upload the image I am getting to my server so that you
will be able to see what I am seeing.

Kind regards,
Marcin Borkowski
--
BPP Marcin Borkowski, ul. Architekt�w 14, 05-270 Marki, Poland
http://www.chembuddy.com, http://www.bpp.com.pl

[Mikhail Kozhevnikov]

Hello Marcin,

It is important to understand that for any structure supplied in the SMILES format the 2d atom coordinates have to be determined prior to rendering. The atom order in the SMILES string is irrelevant. If the automatically determined layout doesn't suit your needs, you can convert the structure to MDL MOLFile, e.g. using indigo-depict itself, than edit the layout manually with an editor of your choice and render the resulting MOLFile.

In the second case the automatic layout is imperfect. We thank you for pointing this out and will consider improving the layout algorithm to handle such cases.
In the meantime, a possible workaround is the same as for the first case: provide the render with coordinates. I found that some other editors make the same mistake, so you might have to adjust the bond orientation manually.

Regards,
Mikhail

--
BPP Marcin Borkowski, ul. Architektów 14, 05-270 Marki, Poland
http://www.chembuddy.com, http://www.bpp.com.pl

--

[m.k.bo...@gmail.com]

Thank you for the answer - I can't say it makes me entirely happy, but at least I know where I am standing.  Manual edits are out of the question, as it is intended as a part of an automatic system, see http://www.chemicalforums.com/index.php?topic=63830.0 to get the idea of the application - SMILES are entered using [smiles][/smiles] tags and rendered automatically. Note it is a test run.

Perhaps the best simple example to explain why the atom order matters is this:

CS

CCS

CCCS

Intuitively I expected them all to be rendered the same way, they are not :(

Still, thank you for your work on the program - being a developer myself I know it is not possible to please everyone, and indigo definitely does things I expected it to do, just in its own way ;)

Regards,

Marcin Borkowski AKA Borek

[Mikhail Rybalkin]

Hello Marcin,

We added an additional check to flip molecule is the first atom is righter than the last one.

You can check the latest indigo-depict here: http://ggasoftware.com/accept?file=indigo-1.1.7%2Findigo-utils-1.1.7b-win32.zip

Let me know how it works. If it works fine we will include this feature in the next release of Indigo.

Regarding your second question, currently we mark only one bond near stereocenter, but in case of 4 bonds we should mark 2 of them.We added this request in our todo list.

Best regards,

Mikhail Rybalkin

[Mikhail Rybalkin]

Hello Marcin,

We have just made a release with all the binaries and libraries included.

Request with 4 bonds near stereocenter is still open.

Best regards,

Mikhail Rybalkin