Interpretation of microarray expression data using ontology browsing

1
Interpretation of microarray expression data
using ontology browsing

• Meeting the needs of a new era

Anders Bresell
2
Motivation
After we have preformed a microarray analysis we
have a set of interesting genes that we want to
know more about.

• The existing knowledge are found in articles and
  databases. But we need a lot of time to skim
  through the information to get a hunch of what
  the knowledge is about.

3
Motivation

• In order to get a first glance of what knowledge
  we have about a set of genes. We need to find a
  way to represent the knowledge in a more
  condensed way.

For this we used Ontologies.
4
What is an Ontology?

• Definition
• An explicit specification of conceptualisation
  Gruber 93

A controlled vocabulary that has specified
relationships between the terms.
5
Why is an ontology so nice?

• It provides a common language of a domain for
  human agents.
• It gives a inter-operability between computer
  systems.
• It enables an improvement of the search and
  knowledge acquisition process.

6
Example of an Ontology
Biological Process
Cell communication
Response to external stimulus
DNA protection
Cell growth and maintenance
Metabolism
Nucleotide metabolism
DNA metabolism
DNA protection
Stress response
DNA protection
7
Ontology 4

• Existing Ontologies

MeSH All published litterature is indexed by the
terms from MeSH.
Gene Ontology Organism independent
ontology describing Molecular function,
Biological process and Cellular component.
8
Information
9
Annotation
Article (in Medline)

• GENE

10
Solution

• Instead of having the information in the form of
  articles and databases we represent it as a set
  of keywords or terms.
• By using ontologies we could specify or
  generalise the terms.

11
Ontology Annotation Tree browser (OAT)

• We built a tool
• Database (holding the annotations and the tree
  structure of the ontology)
• Tree Browser tool (Visualises the ontology)

12
The OAT-tool
Checkbox for terms which are to be included in
the report
Significance of the Annotation
Number of genes annotated with this term
Number of genes below
Number of annotations below
Link to MeSH description
13
Report feature
14
Summary of OAT 1

• Only terms that are relevant for the genes are
  displayed.
• This tool only aims to give some brief
  information about a group of genes.
• The set of genes must undergo further studies.
  The report page directs us to the articles we
  used as source when we made the annotations.

15
Limitations of OAT

• The OAT-tool is totally dependent on the
  annotations. If we have no connections between a
  gene and a term, the tool would not tell us
  anything about the gene.
• We will import more annotations during this
  summer. Then the tool will be a big help in the
  digestion of information.

16
More Annotations

• GeneOntology
• GOA (Swissprot, TrEMBL, Interpro etc.)
• High quality
• Compugene
• Unsupervised!!
• MeSH
• Human Protein (Swissprot, OMIM)
• Text Mining by Ian Dix et al.

17
Modifications

• Change of DBMS from Berkeley DB to MySQL.
• To many annotations.
• To process demanding calculation due to the
  inflexible querying of data.
• Other systems could use the data.
• Makes the updating of data easy.

18
Feedback

• Give me feedback

19
Acknowledgement

• Patrick Lambrix
• Bo Servenius
• Per Broberg
• Thomas Breslin
• Tobias Fändriks