A Data Warehousing Approach for Genomics Data Meta-Analysis - Evolving Application Domains of Data Warehousing and Mining - page 153

and an instance of the relation '' inhibits ''
is detected in the sentence of the given
example,
(ii) Concepts instances are linked by a relation-
ship according to the ontology and grammati-
cal structure: the instances '' p38alpha '' and
`` p38beta '' (subjects of the verb to inhibit)
are linked to the instance '' csaids '' (object
of the verb to inhibit) by the relationship
'' inhibits '',
(iii) A structured annotation describing this in-
teraction is generated as shown on Figure
11.

this information to generate semantic annota-
tions based on GEOnto.
We can distinguish two types of experiments:
so-called public experiments selected by biologists
from the public repositories and so-called local
experiments led directly by the biologist. While
local experiments can be easily annotated using
a dedicated editor to describe them, it is more
difficult to annotate public experiments that are
already described in a variety of ways and not
necessarily in a usable format. That's why we
focused on controlled experiment descriptions:
we chose to annotate experiments selected from
public repositories supporting MIAME. This kind
of repositories allows authors to publish their data
in a usable format. For example, experiments
submitted to GEO are published in MINiML files.
Both MIAME and MINiML are described above
in subsection “Standards and ontologies for micro-
arrays experiments”. We chose to annotate these
files and thus to generate a semantic annotation
of the experiment.
Biologists who submit individual MIAME-
compliant records to GEO provide their data in
three sections: Platform , Samples , and Series .

Annotation of Genomic Experiments

Here is an example of an experiment description
given in natural language:

“The experiment has the accession number
GSE6281 in the GEO repository. It is referred
in a published article (PubmedID 17597826)
describing this experiment: Gene expression
time-course in the human skin during elicitation
of allergic contact dermatitis. The type of study
is described by the keyword allergic response,
and GSM144309 and GSM144311 refer to two
samples that make up this experiment. The
sample GSM144435 concerns a patch test with
5% nickel sulfate taken from a nickel allergic
woman (age range 33-49). The patch test was
exposed for 7h immediately followed by a skin
biopsy...”

• A platform record describes the micro-ar-

ray used in the experiment (e.g. Affymetrix
GeneChips).
A • sample record describes the biological

material under examination and the quanti-
fication measurements derived from those
samples. Sample information is supplied in
two sections: (i) a data table that includes
normalized quantification measurements
from hybridization or SAGE library and

Our purpose is to extract relevant information
describing this type of experiment and to use

Figure 11. Example of MeatAnnot annotation