The document discusses data standards that have been developed for systems biology. Standards facilitate sharing experimental data, allow open-source software development, and are increasingly required for journal submissions. Standards are generally developed by academic groups when an experimental technique becomes established. Examples of standards discussed include mzData for proteomics data, MeMo for metabolomics data, SBML for biological models, and SED-ML/SBRML for simulation experiments and results. Overall, data standards greatly aid computational systems biology by providing frameworks for data sharing and software development.

1. Data Standards for Systems Biology
Neil Swainston
Manchester Centre for Integrative Systems Biology
neil.swainston@manchester.ac.uk

2. Introduction
• Experimental standards
• Proteomics
• Metabolomics
• Enzyme kinetics
• Modelling standards
• Models
• Simulations
• Results

3. Why do we need standards?
• Aids researchers by facilitating management of
experimental data
• Facilitates open-source software development
and interoperability
• Allows data to be shared
• Increasingly becoming a requirement for journal
submissions

4. When are standards developed?
• Standards generally are generated organically
• Not for pioneers
• When an experimental technique becomes
established
• Need for a standard becomes obvious

5. Who develops standards?
• Usually two or more academic groups
• Commercial providers often less enthusiastic
• Often formed by a Working Group
• Proteome Standards Initiative
• Metabolomics Standards Initiative
• “Minimum information required” specification
provided
• Followed by data schema, XML standard

6. MCISB project overview
Enzyme kinetics
Quantitative
metabolomics
Quantitative
proteomics
Model
Parameters
(KM, Kcat)
Variables
(metabolite, protein
concentrations)
PRIDE XML MeMo SABIO-RK
Web serviceWeb serviceWeb service
MeMo-RK
Web service

7. Proteomics
• We wish to store:
• Raw experimental mass spectrometry data
• Protein / peptide identifications
• Protein / peptide quantitations
• Metadata (instrument, search algorithm, user, etc.)

8. Mass spectrometry data
• How do we represent the following?

9. Mass spectrometry data
• The simple approach:

10. Mass spectrometry data
• The simple approach does provide a list of
masses and intensities, but…
• What instrument was used?
• Who ran the instrument?
• What sample was used?
• …etc.
• The simple approach lacks metadata
• Many simple approaches (formats) exist

11. Mass spectrometry data
• The less simple approach: mzData
• Developed by the Proteome Standards Initiative,
2005
• Put together by Working Group of academics and
commercial parties
• Regular meetings, both real and virtual
• Goal: unify the existing “simple” formats into
one
• Support “tagging” with metadata

12. mzData
• http://www.psidev.info/index.php?q=node/80#mzdata
• XML format, includes…
• Peak lists (mz / intensities)
• Experimental protocols
• Admin (Who? When?)
• Instrument details
• etc.

13. Controlled vocabularies
• Use of free text is “dangerous”
• Non-standard, ambiguous terms
• Difficult to match / compare
• Controlled vocabularies
• Collection of standardised terms
• Organised into vocabularies or ontologies
• Ontologies contain controlled terms and relationships
between them (predicates)

14. Controlled vocabularies
• Ontology Lookup Service, EBI

15. mzData

16. Proteomics data
• Proteomics data is not solely mass
spectrometry data
• Sample preparation protocol?
• Peptide / protein identifications?
• Post-translational modifications
• Identification scores?
• To support this, an extension is required
• Extension based on defined set of “minimum
requirements”
• MIAPE

17. MIAPE

18. PRIDE
• Proteomics identifications database
– Both a format and a database
– Centralised, standards compliant, open source, public
data repository for proteomics data
– Query, submit and retrieve proteomics data in
standardized XML formats
– Public version housed at the EBI
– http://www.ebi.ac.uk/pride/

19. PRIDE
• Peptide / protein identifications

20. PRIDE Converter
• User interface
• Usable by biologists
• Interfaces with
Ontology Lookup
Service
• Developed by EBI
• Automatic upload
to PRIDE database

21. PRIDE database

22. Future directions
• PRIDE does NOT hold:
• Protein and peptide quantitations
• New approaches being developed
• mzML – mass spectrometry format, enhancement of
mzData, including support for richer datasets
• mzIdentML – storage of protein and peptide
identifications
• mzQuantML – storage of protein and peptides
quantitations

23. Metabolomics
• We wish to store:
• Raw experimental mass spectrometry (and NMR)
data
• Metabolite identifications
• Metabolite quantitations
• Metadata (instrument, search algorithm, user, etc.)

24. Metabolomics
• Data standard does NOT currently exist
• Core Information for Metabolomics Reporting
• Metabolites Standard Initiative (MSI)
• http://msi-workgroups.sourceforge.net/
• MetaboLights being developed at EBI
• Not many details as yet
• In the mean time…
• MCISB has developed its own repository

25. MeMo
• Metabolomics Model database
• Designed initially for metabolomics data
• SQL / XML hybrid approach
• Holds:
– Experimental meta-data (submitter, lab, date)
– Sample meta-data (including biological source)
– Instrumentation meta-data
– Mass spectra
– Metabolite identifications

26. MeMo

28. MeMo web interface

29. Enzyme kinetics
• How fast does a given reaction occur?
Enzyme
A B
• Determination of kinetic constants which define
the kinetics of the reaction
• Experimental approach: perform kinetic assays

30. Enzyme kinetics
• Many approaches:
– Absorbance
– Fluorescence
– others
• Currently concentrating on absorbance assays
on BMG NOVOstar instrument
• Requirement: determination of KM and kcat for a
given reaction under particular conditions (pH
and temperature)

31. Enzyme kinetics: Michaelis-Menten
• Traditionally, for each assay, initial rate, v is
determined

32. Enzyme kinetics: Michaelis-Menten
• Performing this at various substrate
concentrations allows KM and Vmax to be
determined:

33. STRENDA guidelines
• Standards for Reporting Enzymology Data
• http://www.beilstein-institut.de/en/projects/strenda/
• Specifies…
• Reactants / products
• Enzyme (wild-type, modified, purification, expressed
in
• Experimental conditions (pH, temperature, buffer)
• Instrument, experiment type
• Submitter (contact details)

34. SABIO-RK
• http://sabio.villa-bosch.de/
• Comprehensive collection of enzyme kinetic
constants
• Adheres to STRENDA recommendation
• Harvested from literature
• Searchable web interface

35. SABIO-RK

36. SABIO-RK

37. SABIO-RK

38. BRENDA
• http://www.brenda-enzymes.org/
• Even more comprehensive
• Slightly less well-curated
• Again, searchable web interface

39. BRENDA

40. Other experimental standards
• MIBBI: Minimum Information for Biological and
Biomedical Investigations
• http://mibbi.org/
• Over thirty recommendations for a range of
experimental techniques

41. Modelling standards

42. MCISB project overview
Enzyme kinetics
Quantitative
metabolomics
Quantitative
proteomics
Model
Parameters
(KM, Kcat)
Variables
(metabolite, protein
concentrations)
PRIDE XML MeMo SABIO-RK
Web serviceWeb serviceWeb service
MeMo-RK
Web service

43. MCISB project overview
Enzyme kinetics
Quantitative
metabolomics
Quantitative
proteomics
Model
Parameters
(KM, Kcat)
Variables
(metabolite, protein
concentrations)
PRIDE XML MeMo SABIO-RK
Web serviceWeb serviceWeb service
MeMo-RK
Web service

44. Modelling
• What is a model?
• “An analytic or computational model proposes
specific testable hypotheses about a biological
system”
• Mathematical / computational representation of
a biological system
• May allows computational simulations of the
system

45. Pathway databases
• Building a model often starts with a topological
description of a pathway or pathways
• What reacts with what?
• A number of existing data resources
• Biochemical knowledge, curated from literature

46. KEGG

47. KEGG
Metabolite
Enzyme
Reaction

48. MetaCyc

49. Reactome

50. Simulation tools
• The systems biology community has developed
a strong software infrastructure
• Many tools exist, including simulators
• Several hundred
• How do we link pathway databases to these
simulators?
• A standard: SBML
• Systems Biology Markup Language
• Recently celebrated its 10th
birthday

51. SBML
• XML markup language describing models
• Contains concepts such as…
• compartments
• species (metabolites, enzymes, RNA, etc.)
• reactions
• Similar to pathway databases
• KEGG2SBML tool exists for converting KEGG pathway
maps to SBML files

52. Mathematical SBML
• Also contains concepts allowing simulations
• Many of these driven by experimental work
• Specification of metabolite and enzyme
concentrations
• Specification of kinetic laws and kinetic
parameters
• Parameterised model = pathways + experimental data

53. SBML

54. SBML data resources
• Biomodels.net
• http://www.ebi.ac.uk/biomodels-main/
• Curated collection of biochemical models at EBI
• JWS Online
• http://jjj.mib.ac.uk/
• Also curated
• BUT also includes an online simulator
• You’ll learn more next month…

55. SBML tools
• Hundreds of ‘em (205)
• http://sbml.org/SBML_Software_Guide
• Different goals
• Whole cell / single pathway
• Deterministic / stochastic simulators
• Different platforms / programming languages
• Matrix exists, describing capabilities of each
tool
• http://sbml.org/SBML_Software_Guide/
SBML_Software_Matrix

56. Making SBML models: CellDesigner

57. Other model representations
• CellML
• http://www.cellml.org/
• Larger scale modelling
• Inter-cellular, used in whole organ modelling
• BioPAX
• http://www.biopax.org/
• Similar goals to SBML
• Overlap between “competing” representations
is being reduced
• Regular “COMBINE” meetings

58. MIRIAM
• Minimum Information Required in the
Annotation of Models
• http://www.ebi.ac.uk/miriam/
• Set of guidelines describing how to make
models reusable
• Specify model creator contact details
• Ensure consistent annotation of terms with database
resources
• e.g. use UniProt identifiers for unambigous
identification of enzymes

59. SBML visualisation: SBGN
• Until recently, no standardised way of viewing
models
• Systems Biology Graphical Notation
• Attempts to generate standard “wiring-diagram” for
biological representations

60. Model simulation

61. Model simulation
• Many simulators exist
• How do we tell a simulator what to simulate?
• Simulation Experiment Description Markup Language
(SED-ML)
• Contains concepts…
• Model (what to run the simulation on)
• Simulation (define what to simulate, duration, step-
size)
• Data generation (post-processing normalisation)
• Output (2D plot, 3D plot)

62. Simulation results: SBRML
• Simulation results are data too, and are
represented by SBRML
• Systems Biology Results Markup Language
• Developed by Joseph Dada, et al. (Manchester)
• Structured format for representing simulation
results
• Dada JO, et al. SBRML: a markup language for associating systems
biology data with models. Bioinformatics 2010, 26, 932-938.

63. SBRML

64. Conclusion
• Data standards greatly facilitate computational
systems biology
• Standards exist (and are being continually
developed) for both experimental and modelling
data
• Provides a framework for data sharing and
open-source software tool development

65. Data Standards for Systems Biology
Neil Swainston
Manchester Centre for Integrative Systems Biology
neil.swainston@manchester.ac.uk