Database Reference
In-Depth Information
Figure 2. Transformed distribution comparison between data density in single GSE6475 data (1) and
combined GSE6475-GSE9120 transformed via RMA (2) and CDF (3) on two samples
Tests on Human Skin Allergic Data
not need the individual normalization (per chip
normalization) and per gene normalization. It di-
rectly takes the raw data CEL files of the raw data
set and uses only one time the RMA procedure.
That is the simplest procedure and reduces the
complexity time of computation while returning
similar results.
Furthermore we have led another test: from the
combined GSE6475-GSE9120 data transformed
via RMA and CDF, we have looked for differential
expressed genes betweenACNE and Normal Skin
within GSE6475 data. Table 4 provides results on
differential expressed genes found in these data.
The third and fourth columns give the number of
differential expressed genes found when using
CDF and RMA. The fifth column indicates the
number of differentially expressed genes found
in both cases. We can observe that differentially
expressed genes given by both methods are identi-
cal in a great majority.
In the following section, we present results of
meta-analyses combining Affymetrix data that
are transformed with the RMA-based method
described above.
In this section, we show experimental results ob-
tained when comparing individual analyses with
meta-analyses on gene expression. Our goal is to
observe if individual analysis results are retrieved
when data are combined into a meta-analysis. We
show that the statistical relevance of meta-analyses
is only significant when raw expression data are
combined and then normalized.
We have conducted several tests on meta-
analyses combining the series GSE6281 and
GSE7216 as an example to ensure more reli-
ability in identification of human skin reaction
biomarkers. We detail below results obtained
with a meta-analysis on GSE6281 and GSE7216
series for differential expression search. GSE6281
datasets provide gene expression between time
points (0h, 7h, 48h, 96h) over control subject and
patient subject and GSE7216 datasets provide gene
expression between IL-10 family of cytokines in
reconstituted human epidermis and control subject.
Both are resulting from experiments on the same
Affymetrix chip HG-U133 Plus 2.0 Array. Such
Search WWH ::




Custom Search