Better retrieve SMILES codes from Wikidata instead of Wikipedia

[kelson42]

Wikidata is a structured database and it's therefore easier to parse Wikipedia infoboxes.

Have a look for example to the list of entities including a SMILES code:
https://www.wikidata.org/wiki/Special:WhatLinksHere/Property:P233

[lpatiny]

Thank you for your suggestion that looks very interesting.
Seems that in order to be able this tool we will need some help ... and we have many questions ...

• Could you help us to create the URL that would directly retrieve a JSON containing the article name, SMILES, wikipedia reference
• How often is the list updated ? Is it real time based on wikipedia article modification
• Does this property also look in https://en.wikipedia.org/wiki/Template:Infobox_drug ? I'm not sure based on the definition
• How to do when there is many SMILES (SMILES1, SMILES2, ...) and also comment about the smiles. In some articles we have the smiles for racemate, smiles for R, smiles for S

[kelson42]

I'll do my best, I'm not a wikidata expert...

How to retrieve the list of Molecules with a SMILE code on Wikidata?
https://www.wikidata.org/w/api.php?action=query&prop=linkshere&titles=Property:P233&lhshow=!redirect&lhnamespace=0&lhlimit=500&format=json
Documentation link:
http://en.wikipedia.org/w/api.php?action=help&modules=query

How to retrieve the details about a Molecule?
https://www.wikidata.org/w/api.php?action=wbgetentities&ids=Q132428&format=jsonfm
Documentation link:
https://www.mediawiki.org/wiki/Wikibase/API#wbgetentities

How often is the list updated ?
It's real-time

Does this property also look in Wikipedia?
This is call Wikidata deployment phase2 by us. I'm not sure, but IMO auto. integration between Wikipedia in English and Wikidata is still a discussion topic. On the the Wikipedia in French it works, although this is still not broadly effective https://fr.wikipedia.org/wiki/Wikip%C3%A9dia:Wikidata#Infoboxes_.28phase_2.29

How to do when there is many SMILES?
Not sure to fully understand that question. Maybe this is due to a misunderstanding of what Wikidata is. Wikidata as only one page per molecule and assuming there is only one SMILE per molecule, there is also only one SMILE per molecule wikidata page. You should not have to care then about the infobox in Wikipedia anymore.

[lpatiny]

Thanks for all the answers. We will try to find an internship to improve the wikidata and retrieve not only SMILES but also other interesting parameters and improve our search tool.
Concerning the SMILES problem when multiple SMILES are available the description in
https://en.wikipedia.org/wiki/Template:Chembox (Long parameter list) shows:
Indexed parameters take indexes 'blank'–1–5 (six options). They should have straight input, such as a correct CAS Registry Number. Each parameter can have a comment: |CASNo_Comment=.

Indexed parameters take indexes 'blank'–1–5 (six options). They should have straight input, such as a correct CAS Registry Number. Each parameter can have a comment: |CASNo_Comment=.

Eight base parameters are indexed this way, all identifiers:
CASNo, ChEBI, ChEMBL, ChemSpiderID, DrugBank, InChI, KEGG, PubChem, SMILES, UNII

So in fact we may have up to 6 SMILES and each of them may contain a comment that is sometimes used to described the SMILES or racematate or enantiomerically pure products.

[kelson42]

If there is many SMILES possible for a molecule, then this should be in wikidata too. In all scenarios involving wikidata you should not have to to parse the the Chembox anymore. We can talk this point next time we meet.

[kelson42]

I have asked about the support of multiple SMILES codes here:
https://www.wikidata.org/wiki/Property_talk:P233

I have asked about the lack of elementary physical/chemical properties for chemical components here:
https://www.wikidata.org/wiki/Talk:Q79529