Biomedical Engineering Reference
In-Depth Information
for metabolism, genetic information processing, cellular processes, and organismal
systems; (2) Reactome [95], a free online open-source curated pathway database
encompassing many areas of human biology; and (3) YOUNG [96], encompassing
lists of target genes for known transcription factors. The possibility of obtaining an
overview of the effects of a compound on a cell through transcriptional and pathway
analyses allows not only for a restriction on the number of possible targets (molecules
or complexes) that are affected directly or indirectly by the treatment, but also for
predicting other features, such as the insurgence of resistance to the candidate drug.
Pathway-based microarray analysis methods look for patterns of gene expression
variation in any predefined set of genes. Several tools aiming at this analysis have
been developed. As an example, EuGene Analyser [97], by calculating a
p
-value
using the Fisher exact test (a measure of the significance of a pathway's enrichment
in transcriptionally altered genes), also associates with every output
p
-value a sign
reflecting the relative number of up-regulated genes with respect to down-regulated
genes for each pathway. The use of pathways and gene ontology analyses discloses
important effects of the treatment on the cell. Nevertheless, the identification of a
specific entity targeted by the treatment can be confused with downstream effects.
The use of external data might result in a driving tool for the disclosure of otherwise
unidentifiable information. The procedure lying at the basis of such a comparison,
meta-analysis
, refers to “the quantitative review and synthesis of the results of related
but independent studies” [98], and its application to microarray technologies outputs
does not represent an exception. Previously, meta-analysis was applied to estimate an
average effect of different studies when their results had conflicting conclusions or to
evaluate the degree of agreement of the biological results obtained by various studies
[99,100]. In addition, such an approach was exploited for the identification of com-
mon traits of effects induced by different perturbations. Meta-analysis approaches
have been used widely in different fields [99,101] and key rules have been proposed
[102,103]. With a drug discovery mindset, meta-analysis can take advantage of a
brilliant tool proposed by Hughes et al. in 2000, the Rosetta compendium, a ref-
erence database of expression profiles corresponding to 300 diverse mutations and
chemical treatments in
S. cerevisiae
[85]. The idea driving such a comparison is
that if a newly assayed molecule induces a transcriptional profile similar to the one
induced by the deletion of a nonfundamental gene, the functional gene might be tar-
geted directly or indirectly by the compound being studied. Beltrame et al. proposed
moving the comparison of the transcriptional profile of different treatments from the
gene level to the pathway level [104]. For every condition investigated a pathway
signature is generated, which is a set of descriptors that recapitulate the biologically
meaningful pathways related to the variable (the perturbation) of interest. Whereas
gene-based approaches are generally biased by the analytical procedures employed,
the method based on pathway signatures groups similar samples together successfully
irrespective of the experimental design. By clustering the pathway signatures calcu-
lated for every condition investigated, the similarities among different treatments
can be identified. Use of this approach gave interesting insights for completion of
the functional characterization of PPARalpha, a ligand-activated transcription factor
using a meta-analysis approach on data obtained from several different organisms
