Microarray and Gene Expression

1
(No Transcript)
2
Microarray and Gene Expression Markup
Language (MAGE-ML) Evolution of a Standard
Michael Miller Senior Application
Developer Rosetta Biosoftware I3C Technical
Meeting November 8, 2002
3
Overview
Acknowledgments A World without
Standards Benefits of Standardization The Road
to Standardization Lessons Learned Links
4
Acknowledgments

• Bill Andreopoulos University of Toronto
• Cathy Ball Stanford University
• Doug Bassett Rosetta Biosoftware
• Derek Bernhart Affymetrix
• Alvis Brazma EBI
• Tina Boussard Stanford University
• Steve Chervitz Affymetrix
• Francisco De La Vega Applied Biosystems
• Eric Deutsch ISB
• Michael Dickson LION
• David Frankel IONA
• Jason Goncalves Iobion
• Ken Griffiths LION
• Robert Hubley ISB
• Brandon Hunt Rosetta Biosoftware
• Daniel Iordan Iobion
• Hilmar Lapp Novartis GNF
• Marc Lepage Molecular Mining

• Scott Markel LION
• W.L. Marks Iobion
• Douglas McArthur UC Santa Cruz
• Michael Miller Rosetta Biosoftware
• Helen Parkinson EBI
• Kjell Petersen University of Bergen
• Todd Peterson NCGR
• Angel Pizarro University of Pennsylvania
• Alan Robinson EBI
• Ugis Sarkans EBI
• Martin Senger EBI
• Paul Spellman UC Berkeley
• Jason Stewart Open Informatics
• Marcin Swiatek Imaging Research
• Charles Troup Agilent
• Joe White TIGR
• John Yost National Cancer Institute
• (and many more)

5
A World without Standards

• Proprietary Lock-in
• Customers must use tools specific to the
  technology purchased.
• Allows vendor control over customers' data.
• Continual Development
• Everyone keeps adding or subtracting from a
  moving target.
• What worked one day suddenly stops working the
  next.

6
Benefits of Standardization

• Empowers gene expression data and annotation
  interchange
• Accommodate information from all leading
  industry data formats and gene expression
  platforms.
• Able to convert from existing formats into
  MAGE-ML.
• Exchange data with colleagues.
• Facilitate development of interoperable freeware
  and commercial tools for Conversion,
  Visualization, and Analysis.
• Empowers online supplements to published
  manuscripts
• Provide consistent format for peer review.
• Verify minimum information required for a
  microarray experiment is provided.

7
The Road to Standardization--Beginnings

• OMG Life Sciences Research Technical Committee
  (LSR) issues a Request for Information about Gene
  Expression, January 1999
• Requests organizations to submit requirements
  for interchange of Gene Expression Data.
• EBI, Rosetta Inpharmatics, and NetGenics
  respond.
• EBI replies with what will become the basis of
  Minimum Information about Microarray Experiments
  (MIAME).
• Rosetta Inpharmatics replies with what will be
  the basis of the XML format, Gene Expression
  Markup Language (GEML format) v1.0.
• Microarray Gene Expression Data Group (MGED) was
  established at the International Meeting on
  Microarray Gene Expression Databases, sponsored
  by EBI, November 1999
• Four working groups established MIAME XML
  Format Ontologies and Normalization.

8
The Road to Standardization--Proposals

• LSR issues Gene Expression Request for Proposals
  (RFP), March 2000
• Requirements based on the replies to the Request
  for Information.
• GEML v1.0 XML format intoduced into production
  environment by Rosetta Inpharmatics and Agilent
  Technologies, June 2000
• Launch of Rosetta Resolver system.
• MGED Holds Second Meeting in Heidelberg, May 2000
• Further develops MIAME.
• Initial Microarray Markup Language format
  (MAML.dtd) is created.
• LSR receives the Initial Proposals against the
  Gene Expression RFP, November 2000
• NetGenics submittal based on CORBA interfaces.
• Rosetta Inpharmatics submittal based on GEML
  v2.0 format.
• EBI submittal based on MAML format.

9
The Road to Standardization--Collaboration

• Submitters decide to collaborate
• Submitters meet face to face at MGED3
  conference, March 2001.
• Affymetrix, and other organizations provide
  valuable feedback.
• Submitters adopt OMG Model Driven Architecture
  (MDA) approach
• Start with a blank UML model, eventually named
  Microarray and Gene Expression Object Model
  (MAGE-OM).
• Draw experience from, but do not base the model
  on either GEML or MAML.
• MDA approach allows the submitters to
  concentrate on the domain problem and not worry
  about implementation details.
• Follow up
• Many teleconferences and face to face meetings
  held. Mailing list provides feedback from broad
  community.
• Programming effort is begun after Toronto OMG
  meeting, September 2001, to generate DTD and
  platform code for Java, Perl, and C.

10
The Road to Standardization--Finalization

• Revised Submission becomes Adopted by the OMG
• MAGE-OM v1.0 considered complete, October 2001.
  MAGE-ML.dtd is generated from the model.
• Submission is adopted by the OMG, January 2002,
  and begins its finalization process to work out
  any problems encountered by implementers.
• OMG Finalization Task Force (FTF) and MGED
  Programming Effort
• Finalization Task Force (FTF) formed at OMG to
  handle issues.
• MGED mailing list generates discussion and
  Programming Jamborees implement and test the code
  generated from the model, finding issues and
  potential solutions to those issues.
• Adopted Specification passes OMG vote and becomes
  an Available Specification
• The FTF report is submitted, with an updated
  specification and MAGE-ML.dtd, and accepted by
  the OMG, October 2002.
• MAGE-ML.dtd v1.0 available at http//schema.omg.o
  rg/lsr/gene_expression/1.0.

11
Lessons Learned

• Process
• By going through the OMG adoption process,
  deadlines are setup, ensuring progress will be
  made.
• Requirements and milestones are clearly called
  out by the adoption process.
• Peer review ensures fair treatment.
• Specifications, once adopted, are open to
  members and non-members for comment and use.
• Collaboration
• Working together with fellow submitters lets
  "best of breed" emerge.
• GEML format was strong on those areas that
  benefitted high throughput.
• MAML format was strong in overview of gene
  expression experiments.
• MAGE-OM combines both these ideas.
• All interested parties can contribute, making
  everyone a stakeholder.
• Discourages proprietary approaches by offering
  open alternative.
• Freeware and commercially available tools will
  catalyze conversion, publication, and exchange of
  data.

12
Links

• OMG and Submitters
• Object Management Group www.omg.org
• Life Science Research Domain Task Force
  www.omg.org/lsr
• Rosetta Biosoftware www.rosettabio.com
• Microarray Gene Expression Data Group
  www.mged.org
• Open Source
• Download for reference implementations
• http//mged.sf.net/downloads.shtml
• Discussion group
• mged-mage http//lists.sourceforge.net/lists/lis
  tinfo/mged-mage
• Generating Code
• http//cvs.sourceforge.net/cgi-bin/viewcvs.cgi/mg
  ed