ChemSpider Reactions: Delivering a free community
resource of chemical syntheses
Valery Tkachenko, Colin Batchelor, Daniel Lowe, Ken Karapetyan, David Sharpe and Antony Williams
ACS New Orleans April 2013
Overview
• Motivation• The RSC and chemical reaction data• New sources of chemical reaction data• ChemSpider Reactions: bringing it all together• Experiments with reaction classification• The National Chemical Database Service
Who needs another reaction database?
• Those who cannot afford to license access…• Those who would like to access data that is
not abstracted• Those who might like to contribute data to a
database• Anybody wanting to integrate their systems in
and to pull data out.
RSC and chemical reaction data 1
Graphical abstracting journals:Methods in Organic Synthesis (monthly, 1990 to present)Catalysts and Catalysed Reactions (monthly, 2005 to present)
These constitute a backfile of over 50000 novel reactions
RSC and chemical reaction data 2
RSC and chemical reaction data 3
New sources of reaction data
Daniel Lowe’s PhD thesis (Cantab, 2012) was on extracting reactions from US patent data.We can apply this technology to the RSC Journal archive.
ChemSpider Reactionsbringing it all together
http://csr.dev.rsc-us.org/
WORK IN PROGRESS
Reaction classification 1
Project Prospect has text-mined RSC journal articles for named reactions and molecular processes, annotated according to Creative Commons-licensed ontologies:
See http://rxno.googlecode.com/
Reaction classification 2
Classification of Daniel’s US Patent data
Reaction InChI
To do for reactions what InChI has done for structures•Think online searching•Deduplication and linking
http://www-rinchi.ch.cam.ac.uk/help.html
Reaction InChIEarly work – RInChIs layered on to a few hundred thousand reactions •Not generated for a few 10s of thousands of reactions •Reaction deduplication results differ based on algorithm – GGA software versus RInChI•Under investigation
Other sources
ChemSpider SyntheticPages
•Electronic Lab Notebooks•University repositories
Please send theses
What will ChemSpider Reactions serve?
• Chemical Database Service• Linking back to original
publications/supplementary data• Underpinning other tools e.g. retrosynthetic
analysis (depends on data quality and mapping)
Chemical Database ServiceNational Chemical Database Service for UK academicsIntegrates commercial databases and servicesChemicals, analytical data, prediction algorithmsDevelopment of data repository
ARChem from SimBioSys 1
Synthesis planning tool which performs rule- and precedent-based retrosynthetic analysis back to commercially available starting materials.
ARChem from SimBiosys 2
ARChem from SimBioSys 3
But what about data quality?
• Data validation and curation required• Encouraging participation with
Rewards and RECOGNITION
Manual curation
• Integrated commenting, curating and validation platform across ALL eScience and Publishing platforms
• All integrated to a central RSC profile and feeding the alt-metrics tools
The other kind of RDF(made-up example)
Chemical reactions are unusually well-suited to representation. (Donald Davidson’s event semantics)
_:r1 a obo:RXNO_0000004 ; # Diels–Alder obo:has_participant_ceasing_to_exist _:m1 ;# a diene obo:has_participant_ceasing_to_exist _:m2 ;# an olefin obo:has_participant_starting_to_exist _:m3 .# a substituted cyclohexene_:m1 a <http://rdf.chemspider.com/233000> ._:m2 a <http://rdf.chemspider.com/233001> ._:m3 a <http://rdf.chemspider.com/233002> .
Questions?
E-mail: [email protected], [email protected]