Submitted By: Andrew Tolopko
Adddate: 2008-10-02 12:42:38
Error seems limited to that well/smiles in particular; not a systemic
compound renderer problem.

I'm not one to "see" problems with the smiles string, but John might
see
the problem quickly.  For the sake of time, and the fact that my work
day is over, I'm going to wait for John to take a look on Mon before
re-assuming the initiative to investigate further.

Submitted By: Andrew Tolopko
Adddate: 2008-10-02 12:42:39
Thanks for investigating, John!

Maybe the best we can do is provide an alternate link to, say, the PubChem
renderer for when the Daylight rendered fails?  This would probably require
implementing some additional error reporting capability from our renderer
service.

@

On 16 Jul 2007 09:54:02 -0400, John Sullivan via RT <
reply@rt.med.harvard.edu> wrote:
>
> Apparently Daylight tools don't like the SMILES string, which was
> generated by Open Babel. You can confirm this by trying to display the
> structure by pasting in the SMILES string here:
>
> http://www.daylight.com/daycgi/depict
>
> I found that if you took out the backslash ('\') character preceding the
> "23", Daylight will display it.
>
> It looks like the PubChem structure search tool does not have a problem
> displaying the original SMILES string. You can paste in the SMILES
> string here and click the "Sketch" button and you will see it
properly
> displayed.
>
> http://pubchem.ncbi.nlm.nih.gov/search/search.cgi
>
> Not sure how to handle this further...
>
>

Submitted By: Andrew Tolopko
Adddate: 2008-10-02 12:42:41
> [ant4 - Mon Jul 16 10:29:39 2007]:
> 
> Thanks for investigating, John!
> 
> Maybe the best we can do is provide an alternate link to, say, the PubChem
> renderer for when the Daylight rendered fails?  This would probably
require
> implementing some additional error reporting capability from our renderer
> service.

Well, I could just update the SMILES for that molecule in the database,
but I hesitate to do that, since it is an oddball operation and a more
formal approach is probably called for. But it would fix the problem for
the users!

There should be an extant ticket somewhere for replacing Daylight with
something open source for structure image generation. Which may solve
the problem, if we had a decent image generator, but I was unable to
find one before. There's actually a wiki page to track the issue. There
was one tool, jchemdraw, that I was hopefully waiting on a bugfix for
their image generator.

The PubChem "renderer" isn't actually a rendered, but an image-based
structure editor. As such it would be a little quirky to insert into our
pages. And I don't think there is a programmatic way available to us to
get it to render a certain SMILES string.

We maybe should add a "download molfile" on the Well viewer page, to
help give screeners easy access to the original molfile that everything
else is based on. From that starting point or the SMILES there should be
many options outside Screensaver (eg ChemDraw) for drawing/examining the
structure.

Submitted By: Andrew Tolopko
Adddate: 2009-08-06 17:40:27
Images are now generated and stored as static content.  We're using Daylight
libraries (via Pipeline Pilot) to generate these images outside of Screensaver
proper.  After this point, any images that are not render-able almost certainly
have bad molfile data.