Integrating “Omics” Data for Quantitative and Systems Pharmacology in Translational Oncology - Bioanalysis Advanced Materials Methods and Devices - page 188

discover sets of coordinated differentially expressed genes among pathway members
and their association to a specific biological phenotype. These analyses may provide
new insights linking biological phenotypes to their underlyingmolecular mechanisms,
as well as suggesting new hypotheses about pathway membership and connectivity.
Gene Set Enrichment Analysis (GSEA) [ 78 ]representsoneofthemostcommonly
used knowledge-driven approaches in microarray data analysis.
12.3 Cellular
As biological systems are complex in nature, the interactions between the
components (such as genes and proteins) within a cell form a highly complex
network. The human interactome, which is the study of all molecular interactions
in a human cell, is made by ~20,000 protein-coding genes, ~1,000 metabolites and a
large number of distinct proteins and functional RNA molecules. The total number
of molecular components which serve as the nodes of the interactome easily exceeds
100,000, and the number of functional interactions (edges) between these nodes is
expected to be much larger. To characterize the interactions in the human
interactome, large-scale efforts using the yeast two-hybrid (Y2H) technique have
been employed to systematically identify the interacting human protein pairs
[ 66 , 76 ]. Massively parallel interactome-mapping strategy, Stitch-seq, that combines
PCR stitching and next-generation sequencing technology was recently developed
to facilitate the characterization of human interactome [ 103 ]. Other sequencing
technologies such as ChIP-seq and RNP-seq have been widely used to define
protein-DNA and protein-RNA interactions in biological systems, respectively.
Generating the interactions represent the first step in systems biology [ 40 ]. These
interaction data provide the network topology for a particular biological system.
Network-based analysis and modeling can provide insights into the biological
properties and potential clinical applications in cancer. Computational cellular
models are becoming critical for the analysis of these complex biological systems.
Multiple levels of abstraction can be applied to dissect different biological systems,
ranging from detailed modeling with differential equations, to intermediate topol-
ogy inference using Boolean models and Petri Networks to higher level modeling
by Bayesian networks. Obviously, for model building the more detailed the model
is, the more comprehensive the experimental measurements need to be. These
computational tools and methodologies are facilitating the emergence of systems
biology [ 41 ]. For example, Iadevaia et al. [ 39 ] have developed a computational
approach that integrates mass action modeling with particle swarm optimization to
train on the model against reverse-phase protein array and infer the unknown model
parameters in the insulin-signaling growth factor (IGF-1) signaling network in
breast cancer cell line. Using this computational model, they can predict how
targeting individual signaling proteins alter the rest of the network and identify
drug combinations that inhibit cell signaling and proliferation. This model will be
useful for generating testable hypotheses that could optimize drug combinations
and discover novel pharmacologic targets for cancer therapy.