Biology Reference
In-Depth Information
the condition. The analysis starts by compiling
a list of well-known biomarkers (seed proteins)
related to in
Differential
Analysis
Protein-Protein
Interaction
Network
ammation and prognosis after MI
with the note that none of these biomarkers indi-
vidually predict a positive outcome post MI.
Additional proteins are then added to the list
by identifying proteins with similar functions
as those previously recognized as biomarkers.
The list is then overlaid with protein
e
protein
interaction data to construct a network contain-
ing these seed proteins and all the proteins that
have been identi
Overlay
Datasets
Data
Analysis
ed as a physical interactor to
any one of the seed proteins. This network was
analyzed to identify proteins that the authors
described as
Candidate
Selections
“
critical components and subnet-
works.
This idea of the importance of a protein
being discernible based on a measurement of
several network variables
d
most notably,
whether a protein is a bottleneck protein
30
or
the
”
Independent
Validation
31
of the protein
d
has gained
some traction in recent years. Azuaje et al.
29
clustered these proteins into modules based on
their interconnectivity and attempted to deter-
mine whether the proteins present in a single
module were related to similar biological process
or belonged to the same cellular components.
The authors note that existing biomarkers had
a relatively low connectivity and had a poor
capacity to distinguish between prognostic cate-
gories; however, network analysis identi
“
betweenness
”
FIGURE 1
Typical workflow of integrative network
analysis
e
based discovery and validation of biomarkers. As
discussed in the text,
first differential analysis of the normal
and the disease tissue is performed using standard -omics
technologies. Based on these results, seed genes or proteins
are identi
ed and then combined with protein
e
protein
interaction network to identify subnetworks that have
undergone perturbations. Based on these networks, candi-
date genes are selected for evaluation.
ed
several proteins, such as Inhibitor of Nuclear
Factor Kappa-B Kinase Subunit Epsilon (IKBKE),
TNF Receptor-Associated Factor 2 (TRAF2), and
ubiquitin (UBC) to be highly connected and were
found to have statistically signi
patient are all connected through physical inter-
actions, there is a better chance that those
proteins have a causal role in the disease. There-
fore, their predictive power is greater and the
likelihood of validating these proteins in subse-
quent experiments leading to identi
cant differences
in expression levels between positive and nega-
tive outcome patients as determined by blood
derived microarray experiments. They also
note that the panel of genes identi
cation of
informative biomarker is higher.
An attempt at leveraging protein
e
protein
interaction network data to aid in the identi
ca-
tion of biomarkers was conducted by Azuaje
et al.
29
This study focused on identifying
biomarkers capable of predicting between posi-
tive and negative outcomes post myocardial
infarction (MI) with the additional goal of
improving the mechanistic understanding of
ed through
the network analysis provided a novel prog-
nostic model, which could not have been
revealed by standard analyses.
Another study by Ragusa et al.
32
attempted to
identify biomarkers that predict the likelihood
of relapse following successful
treatment of
Search WWH ::
Custom Search