Biomedical Engineering Reference
In-Depth Information
of the genes in the network. The topology refers to the connectivity structure of the
genes (predictors and predictor set). The interaction describes the dynamical behav-
ior or regulating function of all genes. GRN intervention analyzes the inferred GRN
model to determine which genes can fail, leading to disease, and how to intervene
in the GRN to treat disease. Treatment applies the intervention results to drive new
drug development and direct clinical studies. The research presented in this thesis
focuses on GRN inference and GRN intervention.
The task of inferring a GRN is an arduous one. Because of the complex interaction
of the genes, it is hard if not impossible to construct a single biological experiment
that will yield the complete GRN. Instead, several steps are employed. First, from
biological measurements such as expression microarrays, biologists statistically ob-
serve that a certain subset of genes G are involved in the growth and spread of a
genetic disease. Multiple samples of the gene expression of the genes in G (for sev-
eral diseased and healthy individuals) are taken for comparison. These can also be
in the form of time course data (where expression of the genes in G are taken for
the same individual, over a sufficiently long duration). Time course data is generally
not readily available, however. From the gene expression data, logic techniques can
be utilized to i) find the support or predictors for each gene g i
G , and ii) infer the
function of the GRN. To validate the GRN obtained in this manner, biologist can
perform targeted experiments to verify specific gene interactions within the GRN.
Often, pathways (or portions) of the GRN are known, from targeted experiments
that have already been conducted by biologists in the past. By curating the results of
several such (often independently conducted) experiments, some GRNs have been
inferred with reasonable confidence. This information can be used to verify results, or
used as additional inputs to constrain the state space of our logic synthesis methods.
Once the GRN for a genetic disease is known, one main area of interest is to
understand how genetic disease arises from the GRN and how to intervene in the
GRN to treat diseases. In the case of genetic diseases, genes are mutated or damaged
leading to signaling failure in the GRN. The problem then becomes how to identify
which genes are responsible for a particular disease and how to design drugs to correct
the behavior of these genes. If the specific effect of candidate drugs on particular
genes is known, another problem of interest is to find the best set of drugs which
correct the GRN behavior of a diseased organism. Both problems can be cast as
instances of another logic synthesis method called automatic test pattern generation
(ATPG).
1.5
Logic Synthesis
The Boolean Network [ 12 ] model for GRNs is a finite state machine (FSM), which
provides the motivation for applying established and efficient techniques and algo-
rithms from the field of logic synthesis. In computer circuit design, logic synthesis
is the process of converting a high-level specification into an optimized logic gate
representation. Logic synthesis techniques can be further split into methods, which
Search WWH ::




Custom Search