Analysis of Regulatory and Interaction Networks from Clusters of Co-expressed Genes - Clustering Challenges in Biological Network - page 68

After establishing the fact that each individual cluster appears to have a set
of co-functional genes, the transcription factors that could possibly co-regulate
these genes can be identified. These transcription factors therefore may provide
points of possible transcriptional intervention to alter significant biological pro-
cesses. Therefore it may be possible to alter the inflammatory response without
changing the immune response, or the metabolic responses. It is our hypothesis
that if a transcription factor can be associated with a primary ontology and if these
transcription factors are associated with a large majority of genes in the given clus-
ters, then it could provide a possible point of intervention. A list of these possible
transcription factors are given in Table 1.
The transcription factors that comprised the highly connected nodes, i.e., their
around found to be regulating all the clusters, can be broken down into two sets,
the set of transcription factors that are highly connected throughout the genome
and the set of transcription factors that are highly connected only to the genes
that were selected as relevant. For instance, transcription factors such as STAT
5 and STAT 6 were selected as highly connected. However, these transcription
factors are also highly represented in a set of randomly selected genes belying
their important biological role, and while these transcription factors undoubtedly
play a role in the corticosteroid response, they do not represent valid points at
which the corticosteroid response can be mediated due to their widespread ef-
fects on many other systems, as evidenced by their presence in every functional
and co-expressed cluster. The other transcription factors such as CDX (Caudal-
type homeodomain protein), AP2-Alpha (activating enhancer binding protein 2),
USF(Upstream Stimulating Factor), and PAX4(Paired Box Gene 4) represent hubs
that are highly connected to the selected genes and may present themselves as
possible targets to mediate the response. Coupling the biological information ob-
tained via iHOP [64], it becomes possible to ascertain the role that each hub may
play. These four transcription factors were specifically chosen for further analysis
from the set of hub genes because additional information as to the up or down-
regulation of the proteins that code for these transcription factors is available in
the form of mRNA expression data. Therefore, it becomes possible to rationalize
their biological activity with their observed activity and the consequence upon cor-
ticosteroid response. For instance, the down-regulation of USF may point to the
decrease in lipid and glucose metabolism by the liver [65], leading to the observed
increase in level of circulating free fatty acid and glucose in the bloodstream after
an infusion of corticosteroid in a rat. The up-regulation of AP2-Alpha hints at a
possible mechanism by which corticosteroids may suppress cellular proliferation
given its role in suppressing the growth of malignant tumors [66].
Two of the transcription factors that are highly connected in the network pose