chebi and genome scale metabolic reconstructions
TRANSCRIPT
ChEBI and genome-scale metabolic reconstructions
Neil SwainstonManchester Centre for Integrative Systems Biology
2nd ChEBI User Group Workshop, EMBL-EBI, Hinxton, UK23 June 2010
Introduction
• Genome-scale metabolic reconstructions
• Why ChEBI?
• Kinetic modelling and experimental data
• Conclusions
Genome-scale metabolic reconstructions
Genome-scale metabolic reconstructions
• Systems biology approach• Network map of metabolism
• Collection of metabolic reactions
• Computer-readable• Uses
• Metabolic engineering• Biological discovery: gene annotation• Phenotypic behaviour: gene essentiality• Bacterial evolution• Knowledge base: computational representation
Systems Biology Markup Language
• SBML• Established community standard
• ~10 years old• >500 citations• >180 software packages
• XML language marking up:• chemical species (metabolites, enzymes, genes)• reactions• compartments• more…
Systems Biology Markup Language
Limitations
• What does Cytoplasm, D-glucose, YGL253W mean?• Same as cyt, glc, Hexokinase 2?
• How can we share, compare, merge, parameterise models?• Some “standard” is required
• Minimal Information Required In the Annotation of Models• MIRIAM
MIRIAM
• Standardised format for model annotation• Subset of Resource Description Framework (RDF)
• Mapping of model constituent to external data resource
• For metabolites, we use ChEBI:• D-Glucose is CHEBI:17634
MIRIAM annotation
<species metaid="id_1" name="D-Glucose">
<annotation>
<rdf:RDF>
<rdf:Description rdf:about="#id_1">
<bqbiol:is>
<rdf:Bag>
<rdf:li rdf:resource="urn:miriam:obo.chebi:CHEBI:17634"/>
</rdf:Bag>
</bqbiol:is>
</rdf:Description>
</rdf:RDF>
</annotation>
</species>
For example…
• Genome-scale SBML model of yeast metabolism• MIRIAM compliant
– All 2153 molecules have unique database references– Community standards have been followed– Should be entirely unambiguous for third party users– Should be usable in third party tools
Why ChEBI?
• Could also use KEGG, PubChem, HMDB?• Occasionally we do…
• Excellent search facility• Both web and web service
• Web service allows us to develop automation tools• SBML Annotation Wizard
SBML Annotation Wizard
• Principle:– Iterate through model– Submit each species to an appropriate web service
search facility (ChEBI or UniProt)– Search names and synonyms– Prompt user for confirmation– Update model with annotation
SBML Annotation Wizard
• Can also be configured to search with chemical formulae• Other tools exist with similar functionality
Why ChEBI?
• Good coverage, but “holes” exist
• Submission tool• During recent work, molecules have been submitted
to ChEBI• Rapid turn around
Why ChEBI?
• Fast search facility• Can be exploited to work “in real time”• Autofill facility
• Rather than annotate models after generation, use annotation to drive their generation
• SuBliMinaL
SuBliMinaL
Autofill
• Useful elsewhere• Used in experimental data metadata collection• Excel plugin?
Why ChEBI?
• Comprehensive web service facility• Goes beyond names• Can retrieve pretty much everything from web
service that is available through web interface
Reaction balancing
• Access to charge and formula
• Can automate model checking process• Correct protonation state• Correct stoichiometry• Missing cofactors
• Application written• Useful (and necessary) for genome-scale
reconstructions
Reaction balancing in SuBliMinaL
Interfacing with cheminformatics
• ChEBI annotations allows us to think of networks cheminformatically as well as mathematically• Programmatic access to InChI strings• InChI strings can be computationally interpreted• Automated network gap filling?• Similarity searching of “danglers” within network
Kinetic modelling project overview
Enzyme kineticsQuantitativemetabolomics
Quantitativeproteomics
SBML Model (single pathway)
Parameters(KM, kcat)
Variables(metabolite, proteinconcentrations)
PRIDE MeMo SABIO-RK
Web serviceWeb serviceWeb service
MeMo-RK
Mapping experimental data to models
• Relatively easy
• Model components and data annotated with same terms• ChEBI and UniProt
• ChEBI acts as “glue”
Conclusions
• Metabolic reconstructions (and experimental data) require annotation
• ChEBI provides comprehensive resource for representing metabolites• Also welcome new submissions
• ChEBI provides excellent web services, allowing the development of automated tools• Synonyms, charge, formula, InChI string• Also mass, hierarchical relationships
Thanks…
No Thanks…
Many Thanks…
ChEBI and genome-scale metabolic reconstructions
Neil SwainstonManchester Centre for Integrative Systems Biology
2nd ChEBI User Group Workshop, EMBL-EBI, Hinxton, UK23 June 2010