Comments on: Dictionary Lookups and Optical Structure Recognition Versus Structure Drawing. Which is Less Error Prone?

By: ChemSpider Blog » Blog Archive » Curators Perform Heroic Duties. They Should be Celebrated!

Thu, 04 Oct 2007 16:48:39 +0000

[...] Then I commented on there are better ways to ensure the quality of structure drawings than redrawing them…specifically dictionary look-up and optical structure recognition. [...]

By: ChemSpider Blog » Blog Archive » CLiDE and more complexity. ChemSpider has Agents and EyeBalls

Thu, 04 Oct 2007 03:46:09 +0000

[...] I wrote about the potential advantages of using Dictionary Look-Up and Optical Structure Recognition over the “copying of [...]

By: Antony Williams

Antony Williams — Wed, 03 Oct 2007 00:02:06 +0000

Liquidcarbon has acknowledged my observation that a hydroxyl group was dropped during the redrawing of Discodermolide.

“That’s true, my bad. I was copying from the paper (Org. Proc. Res. Dev. 2004, linked in the post). Thanks!”

To reiterate the purpose of this blog post (relative to all the stuff made up about it http://wwmm.ch.cam.ac.uk/blogs/murrayrust/?p=652) – if you have the chemical name look up the structure on a database if you can. If you don’t have the name-structure associations available then use optical structure recognition software as a good starting point and tweak the structure.

By: Antony Williams

Antony Williams — Tue, 02 Oct 2007 23:40:03 +0000

An important detail I missed last night: CLiDE was initially developed by Peter Johnson’s research group at the University of Leeds. It is marketed and supported by Simbiosys. CLiDE Pro is developed by Keymodule Ltd (based in Leeds) and marketed in North America by Simbiosys.

By: Antony Williams

Antony Williams — Tue, 02 Oct 2007 23:34:41 +0000

Just for completeness the Patent regarding chemical structure recognition from scanned structures issued by Boyer is at http://www.patentstorm.us/patents/5345516.html

By: Unilever Centre for Molecular Informatics, Cambridge - petermr’s blog » Blog Archive » Can chemical structures be right or wrong?

Tue, 02 Oct 2007 07:08:35 +0000

[...] ChemSpider Blog Â» Blog Archive Â» Dictionary Lookups and Optical Structure Recognition Versus Struc… Says: October 2nd, 2007 at 5:48 am e[â€¦] Luqidcarbon has put up a recent blog posting about the speed by which he/she can draw structures in ChemDraw and asked for challengers. PRM has commented in Chemical SpeedDrawing. The challenge is outlined belowâ€¦ [â€¦] [...]

By: Antony Williams

Antony Williams — Tue, 02 Oct 2007 05:19:30 +0000

The full details of the processing are as follows:

The recognition process took about 10 seconds for the three structures on a PC with a Pentium4 3GHz processor.

The results are:
– THC: H at the wedged bond is unrecognized.
H at the dashed wedged bond is not connected to the
structure because the dashed bond is not recognized well.

– Penicillin: CT is fine.
Only mistake is that the wedgeness of the wedged bonds is
not recognized.

– Discodermolide: 1 mistake in CT: dashed bond at the atom label
of OCONH2 is not recognized; therefore OCONH2 is not
connected to the structure. Again, solid wedges are not recognized.

All of the edits would be EASY to make.