ANNOUNCE: Chempound repository now live
Peter Murray-Rust
http://quixote.ch.cam.ac.uk/content/compchem/index.html
It is searchable by SMILES (try c1ccccc1Cl for Chlorobenzene which gives 6 results)
Sam has done a fantastic job and will be presenting this at OR11 today in Austin.
**NOTE IT IS A PROTOTYPE AND HAS NOT BEEN DEBUGGED. SO THERE MAY BE ERRORS OF EXTRACTION, TRANSCRIPTION, ETC. ** (Sam, can you add a notice to the splash page making this clear).
We will discuss this at the Quixote meeting later this week (Thurs/Friday). Please join us.
Wish list and design. There are now MANY things people will want to do. Ongoing activities include:
* extraction of more properties (esp thermochemistry and spectra)
* addition of NWChem results
* global compchem dictionary
* conventions for properties
We are open to donations of collections, initially in Gaussian or NWChem format, but will have to prioritize them. We would be delighted to work with people wishing to clone Chempound/Quixote for public display of molecules.
There are some clear imperatives:
* annotations (we have ideas based on blogging)
* geometrical searches (initially bond lengths)
The best way to get things into the system is to join the activity and contribute. There are tutorials on how to write parsers, dictionaries, etc.
P.
--
Peter Murray-Rust
Reader in Molecular Informatics
Unilever Centre, Dep. Of Chemistry
University of Cambridge
CB2 1EW, UK
+44-1223-763069
Egon Willighagen
thanx for the new version!
On Wed, Jun 8, 2011 at 9:11 AM, Peter Murray-Rust <pm...@cam.ac.uk> wrote:
> **NOTE IT IS A PROTOTYPE AND HAS NOT BEEN DEBUGGED. SO THERE MAY BE ERRORS
> OF EXTRACTION, TRANSCRIPTION, ETC. ** (Sam, can you add a notice to the
> splash page making this clear).
I searched for InChI=1S/CH4/h1H4 and also got hits for CH2
calculations... that could make sense, and may be intended, but you
may want to clarify that the InChI linked to entries does not
necessarily reflect the structure studied in that entry, like here:
http://quixote.ch.cam.ac.uk/content/compchem/bangor/anna_351_400/365/index.html
I have added the RSS feed to this FriendFeed group:
http://friendfeed.com/openchemicaldata
(only three items showed up in the feed, which I had to delete, as the
link was wrong)
Looks really great!
Egon
--
Dr E.L. Willighagen
Postdoctoral Researcher
Institutet för miljömedicin
Karolinska Institutet (http://ki.se/imm)
Homepage: http://egonw.github.com/
LinkedIn: http://se.linkedin.com/in/egonw
Blog: http://chem-bla-ics.blogspot.com/
PubList: http://www.citeulike.org/user/egonw/tag/papers
Sam Adams
On 08/06/2011 08:30, Egon Willighagen wrote:
>
> I searched for InChI=1S/CH4/h1H4 and also got hits for CH2
> calculations... that could make sense, and may be intended, but you
> may want to clarify that the InChI linked to entries does not
> necessarily reflect the structure studied in that entry, like here:
>
> http://quixote.ch.cam.ac.uk/content/compchem/bangor/anna_351_400/365/index.html
I'll try to debug that. The InChI library can automatically 'adjusts'
hydrogen counts sometimes, depending on how you formulate the input...
> I have added the RSS feed to this FriendFeed group:
>
> http://friendfeed.com/openchemicaldata
>
> (only three items showed up in the feed, which I had to delete, as the
> link was wrong)
It's an atom archive paged feed (http://tools.ietf.org/html/rfc5005).
If you follow the link@rel="prev-archive" records you can retrieve the
entire history of the repository. I've got a java library that will
help with that, if its useful.
> Looks really great!
Thanks!
Sam
Egon Willighagen
> I'll try to debug that. The InChI library can automatically 'adjusts'
> hydrogen counts sometimes, depending on how you formulate the input...
This ties in a bit on how people can leave comments with entries...
(just thoughts :)
>> I have added the RSS feed to this FriendFeed group:
>>
>> http://friendfeed.com/openchemicaldata
>>
>> (only three items showed up in the feed, which I had to delete, as the
>> link was wrong)
>
> It's an atom archive paged feed (http://tools.ietf.org/html/rfc5005). If
> you follow the link@rel="prev-archive" records you can retrieve the entire
> history of the repository. I've got a java library that will help with
> that, if its useful.
Sounds cool! Did not know Atom could do that... it's not going to help
with FriendFeed, though. Might it's good for pulling down the whole
archive, but there are better ways of doing that (e.g. create an index
with SPARQL first :)
Sam Adams
> On Wed, Jun 8, 2011 at 1:12 PM, Sam Adams<se...@cam.ac.uk> wrote:
>> I'll try to debug that. The InChI library can automatically 'adjusts'
>> hydrogen counts sometimes, depending on how you formulate the input...
> This ties in a bit on how people can leave comments with entries...
> (just thoughts :)
My current thoughts are towards a 'trackback' mechanism, like blogs have...
>> It's an atom archive paged feed (http://tools.ietf.org/html/rfc5005). If
>> you follow the link@rel="prev-archive" records you can retrieve the entire
>> history of the repository. I've got a java library that will help with
>> that, if its useful.
> Sounds cool! Did not know Atom could do that... it's not going to help
> with FriendFeed, though. Might it's good for pulling down the whole
> archive, but there are better ways of doing that (e.g. create an index
> with SPARQL first :)
The advantage of atom archive feeds is that you can poll them, and just
receive the updates... only the 'current' document changes; once a page
has moved to the archived state it should never change (though this
repository will probably get 'rebooted' a few times before things
stabilise...
Sam
Egon Willighagen
> My current thoughts are towards a 'trackback' mechanism, like blogs have...
Yes, sounds neat! Quite like the article level metrics at PLoS...
> The advantage of atom archive feeds is that you can poll them, and just
> receive the updates... only the 'current' document changes; once a page has
> moved to the archived state it should never change (though this repository
> will probably get 'rebooted' a few times before things stabilise...
:)
Is this approach also used in CrystalEye?