CCL: extract chemical information from PDF tables

The quality is comparable to OCR - optical character recognition - and I've always called this method "Chemical OCR." If you're familiar with OCR, it makes a LOT of mistakes. For example, CLiDE almost always fails to recognize iodine, which it interprets as a methyl group with implicit carbon and hydrogen. Or vice versa -- if your methyl group happens to be vertical, it often gets misinterpreted as an iodine atom. Rings sometimes end up "broken", i.e. CLiDE interprets a 6-membered ring as a chain with 6 atoms arranged in a hexagon. So it's not perfect. As with text OCR, you have to read the result and correct it manually.

I found that using CLiDE, whether in one-at-a-time form or batch form, requires manually comparing the original with the chemical-OCR-interpreted result (with text OCR you can usually skip referring back to the original since you know all the words in your head. But with chemical OCR, the two examples I gave above are obvious upon inspection, but many other errors are not). The best method to inspect is to assign names to the resulting structures using some automatic structure-to-name converter. That exposes any subtle issues like an atom being "close" to a methyl group when it should be an atom connected by a bond. Compare the automatically generated name to your visual inspecction of the original, and the process is pretty quick.

On 5 April 2012 12:42, Alex Allardyce aa|a|chemaxon.com <owner-chemistry-,-ccl.net> wrote:

Sent to CCL by: Alex Allardyce [aa : chemaxon.com]
ChemAxon supports text and scanned pdf (and doc, ppt, pptx etc). It is integrated throughout our technology but probably easiest to try out in MarvinView, (free for the desktop), just 'open' the pdf (or doc or ppt etc) and all extracted structures are shown. You also access all of the functionality through the API/command line as well as KNIME and Pipeline Pilot nodes.

Cheers
Alex

On Tue, Apr 3, 2012 at 17:55, Brian Bennion bennion1=-=llnl.gov
<owner-chemistry[*]ccl.net> wrote:

Sent to CCL by: "Brian Bennion" [bennion1=llnl.gov]
Hello,

Does anyone have/know of code to parse pdf tables for chemical structure and activity data?

Searching the web did not result in much so I may not be searching with the correct terms. One interesting hit was the clide code from simbiosis.

Has anyone used this for pulling structures out of pdf files?

I want to populate a repository with chemical structures and annotate the entries with the activity data given in an associated table located in the same pdf document.

Thanks
Brian>

-= This is automatically added to each message by the mailing script =-
E-mail to subscribers: CHEMISTRY-,-ccl.net or use:
http://www.ccl.net/cgi-bin/ccl/send_ccl_message

E-mail to administrators: CHEMISTRY-REQUEST-,-ccl.net or use
http://www.ccl.net/cgi-bin/ccl/send_ccl_message
http://www.ccl.net/chemistry/sub_unsub.shtml

Before posting, check wait time at: http://www.ccl.net

Job: http://www.ccl.net/jobs Conferences: http://server.ccl.net/chemistry/announcements/conferences/

Search Messages: http://www.ccl.net/chemistry/searchccl/index.shtml
http://www.ccl.net/spammers.txt

RTFI: http://www.ccl.net/chemistry/aboutccl/instructions/