A Data Warehousing Approach for Genomics Data Meta-Analysis - Evolving Application Domains of Data Warehousing and Mining - page 145

Table 7. Comparison of differentially expressed genes elicited by an individual analysis on GSE6281
and a meta-analysis combining raw CEL data of GSE6281-GSE7216 series

Comparison Individual analysis Meta-analysis Gene in common

control -7h/0h 219 280 142

control -48h/7h 15 42 8

patient -7h/0h 223 308 164

patient -48h/7h 3069 4655 2798

Table 8. Comparison of differentially expressed genes elicited by an individual analysis on GSE7216
and a meta-analysis combining raw CEL data of GSE6281-GSE7216 series

Comparison Individual analysis Meta-analysis Gene in common

IL19-control 82 90 77

IL20-control 207 247 189

IL22-control 606 701 580

IL24-control 363 429 349

IL26d-control 0 2 0

KGF-control 224 254 202

IFNg-control 253 291 242

IL1b-control 214 267 199

exPreSSIon dAtA, SynthetIc
dAtA And MetAdAtA

comparison and select common probe sets only
for data integration (Jiang2004).
In summary , in this section, we have intro-
duced issues of gene expression meta-analyses
that concern mainly the integration of heteroge-
neous data. We have observed that their statistical
relevance is only effective when data are coming
from very identical studies and micro-arrays tech-
nologies, thus motivating the design of a tool like
AMI that allows to leading meta-analyses at a more
synthetic level. Indeed the AMI framework will
allow either to drive standard meta-analyses on
gene expression data (as described in this section)
or to start from preliminary results (synthetic data)
provided by individual statistical analyses and to
mine them for extracting potentially interesting
knowledge. In the next section, we give details on
the representation and the storage of these various
data types in AMI.

This section is dedicated to the representation
and the storage of raw data on gene expression
intensity (first subsection), to synthetic or refined
data that result from previous analyses on raw
expression data (second subsection) and to meta-
data that are one important underlying support to
semantic capabilities in AMI (third subsection).
These data are stored in the AMI knowledge base
as detailed below.

representation of expression data

One goal of the AMI data warehouse is to give
the capability to store raw datasets (represented
by (4) in Figure 1) that contain gene expression
intensities and to explore more deeply original
experimental results with further so called in silico
individual or meta analyses. These data may be