Content, Format, and Standards in Genomics Scale Data

1
Content, Format, and Standards in Genomics
Scale Data

• The ILSI EBI Collaboration
• Wm. B. Mattes, PhD, DABT

2
Outline

• Why do we need a database for toxicogenomics
• How is it envisioned that this will be developed
• What are the issues for such a database
• Who is involved in such development
• The ILSI EBI Collaboration

3
Traditional Biology
One tree at a time
4
Omic Biology
Forests and Mountains
5
Challenge of Genomics

• Its the informatics, period!
• And its awfully tempting to take shortcuts!

6
Why do we need a database?

• Volume of data
• Traditional endpoints per animal
• lt20 histopathology observations
• lt10 gross measurements (e.g. weights, food)
• lt25 serum measurements
• lt10 urinalysis measurements
• Genomic endpoints per animal
• 5,000-10,000 transcripts !!!

7
Why do we need a database? (cont)

• Influence of technology details
• Influence of probe sequence
• Many genes are alternatively spliced such
  events may not be detected unambiguously by a
  microarray

8
Influence of Probe Sequence
Most arrays target this region of the mRNA!
9
Why do we need a database? (cont)

• Influence of technology details
• Influence of probe sequence
• Many genes are alternatively spliced such
  events may not be detected unambiguously by a
  microarray
• For cDNA arrays, probes may hybridize to more
  than one sequence
• A database that captures probe sequence is
  required to resolve discrepancies through
  automated bioinformatics

10
How are databases being developed?

• Microarray Gene Expression Data Society - MGED
  Society
• MIAME - Minimum Information About a Microarray
  Experiment
• the minimum information that should be reported
  about a microarray experiment to enable its
  unambiguous interpretation and reproduction
• Essentially, what should go into the database

11
How are databases being developed?

• MIAME Basic Areas
• Experiment Design
• Samples used, extract preparation and labeling
• Hybridization procedures and parameters
• Measurement data and specifications
• Array Design

12
How are databases being developed? (cont)

• MGED Society
• MAGE
• Programming conventions and data structures to
  communicate Microarray Gene Expression data
• MAGE-OM Object Model
• MAGE-ML Markup Language
• Essentially, how the data is exchanged/ how the
  database is constructed

13
How are databases being developed? (cont)

• MGED Society
• Ontology working group
• Ontologies provide a vocabulary for representing
  and communicating knowledge about a
  topic,allowing interpretation and use by
  computers
• MGED Ontology will provide standard terms for the
  annotation of microarray experiments, allowing
• structured queries
• unambiguous descriptions of experiments

14
How are databases being developed? (cont)

• MGED Society
• Data Transformation and Normalization Working
  Group
• Standards for recording how microarray data are
  transformed and normalized.

15
What are the issues for a toxicogenomics database?

• Scope of the ILSI effort
• Genotoxicity Group
• 10 array platforms
• 11 compounts
• gt2 time points, up to 10 doses / compound
• Nephrotoxicity Group
• 6 array platforms
• 3 compounds, 260 animals

16
What are the issues for a toxicogenomics database?

• Scope of the ILSI effort
• Hepatotoxicity Group
• 8 array platforms
• 2 compounds, 144 animals
• 2 in-life studies / compound
• ALL Groups
• Analysis of each sample at multiple sites

17
What are the issues fortoxicogenomics databases?
(cont)

• Traditional toxicology endpoints are not
  currently covered by MAGE, MIAME, or the MGED
  Ontologies!
• Organ weights
• Clinical pathology
• Histopathology
• Etc

18
What are the issues for toxicogenomics databases?

• Traditional toxicology endpoints are not
  standardized in nomenclature
• Clinical pathology/chemistry
• AACC
• IUPAC
• Histopathology
• STP
• WHO/IARC/RITA
• NACAD
• SNOMED
• NTP, TDMS Database Pathology Code Table

19
Who is involved in database development

• Private Companies
• Genelogic, Iconix, Curagen
• MSU - dbZach
• NIEHS - CEBS
• NCTR - ArrayTrack
• ILSI - EBI

20
ILSI-HESI and EBI collaboration

• Establishment of database for toxicogenomics data
• Capture, store and analyse gene expression data
  produced from many different toxicogenomic
  experiments, conducted in several different
  laboratories worldwide by the ILSI-HESI members
• Interrogate the gene array data integrating
  information from genomic, experimental and
  toxicological domains
• Gain knowledge of possible links between gene
  expression changes and toxicological endpoints

21
ILSI-HESI and EBI collaboration

• Aims of the database and tools
• Provide a way to integrate the different domains
• Control the annotation to achieve data
  harmonization
• Centralize the information to ease data access
  and data sharing
• Improve array annotations as the genome
  assemblies are released
• ALLOW data comparison

22
ILSI-HESI and EBI collaboration

• Main challenge
• Get internally consistent data to allow
  comparability among the experiments and run
  complex queries across and within domains
• Note Experiments conducted in 40 different
  sites, using different array platforms and
  terminologies, measuring parameters with
  different units and storing information in
  different format !

23
ILSI-HESI and EBI collaboration

• Simple question
• Does gene X expression goes up after treatment
  with compound Y with biological endpoint Z in
  experiments from ILSI-HESI members A and B ?
• Not simple question
• Which are the most reproducible gene expression
  changes (and the quantitative measure of this
  reproducibility) for all experiments on the rat
  arrays, with biological endpoint X, and which
  functional category these genes belong to and
  which are the human homologues ?

24
MIAME/Tox

• An international effort aiming to
• Share expertise
• Encourage harmonization
• Promote standardization initiative
• A call for community participation!

25
MIAME/Tox objectives

• Standard contextual information
• Establish worldwide scientific consensus on the
  minimal information descriptors for array-based
  toxicogenomics experiments
• Data harmonization
• Encourage use of controlled vocabularies for the
  toxicological assessments
• Data integration and data sharing
• Link data within a study
• Link several studies from one institution
• Exchange datasets among institutions
• Data storage
• Facilitate development of MIAME/Tox compliant
  data management softwares and databases
• - ArrayExpress _at_ EBI and CEBS _at_ NIEHS-NCT

26
MIAME/Tox document

• Promote standard contextual information
• Defining the core common to most experiments
• - Minimum/sufficient information
• Structured information
• Promote data harmonization, data capture and
  communication
• MIAME/Tox is based on MIAME
• Focus on toxicological domain
• Sample treatment and conventional toxicology
  information
• - Clinical pathology, pathology, histopathology

27
MIAME/Tox document

• Available at the MGED Society and ILSI-HESI web
  sites
• Circulate for consensus
• Toxicogenomics, pharmacogenomics and
  ecotoxicogenomics communities
• - Regulatory bodies
• MGED Meeting (AAAS, Denver, Feb 2003 MGED6,
  France, Sept 2003)
• - Toxicology societies (SOT Meeting, Salt Lake
  City, March 2003)
• Review and publish
• Work closely with the MGED working groups
• Ontology working group
• Identify controlled vocabularies for
  toxicological metadata

28
Data Input As a Key Step

• Capture data in a standard manner
• Tox-MIAMExpress
• Store information domains in database
• ArrayExpress
• Compare/query across and within domains

29
Tox-MIAMExpress
30
Tox-MIAMExpress

• Array designs
• A set of procedures for formatting the array
  design information into a standard referencing
  format (ADF)
• A set of procedure to re-annotate or up date the
  array designs via a link to another database at
  EBI (EnsMart)

31
Tox-MIAMExpress

• Experiment
• Experiment design, quality controls,
  publications
• Sample source and treatment
• Conventional toxicology tests data
• Microarray hybridizations data

32
Tox-MIAMExpress
33
Tox-MIAMExpress
34
Tox-MIAMExpress
35
ILSI-HESI and EBI collaboration

• Status
• Interface and database infrastructure developed
• Data input ongoing

36
Acknowledgments

• Microarray Informatics Team at EBI, in particular
• Alvis Brazma (Team Leader and MGED Society
  President)
• Susanna-Assunta Sansone
• Philippe Rocca-Serra (Data Management)
• NIEHS-NCT and NTP
• ILSI-HESI EBI Steering Committee
• ILSI-HESI Genomics Committee