Biology Reference
In-Depth Information
we can reconstruct genome-scale metabolic models using as much
available information as possible. A good example of this approach has
been reported by Segre et al. [81], who suggested an automated process
to construct stoichiometric models from annotated genomes. Obviously,
data contents and their quality in the metabolic databases are impor-
tant in constructing reliable metabolic models. Several excellent
metabolic databases are available (table 7.1) and are expected to evolve
as an integrated one, as demonstrated for the BioSilico database.
Once the valid model is constructed, the system's behavior under
a particular experimental situation can be better predicted by system-
atic perturbations encompassing genetic to environmental alterations.
Accordingly, in silico experiments of the system through metabolic
modeling can provide crucial information on cellular behavior under
various genetic and environmental conditions, thereby giving rise to
a multitude of strategies for strain improvement. For example, using
the genome-scale in silico model, constraints-based flux analysis can
be carried out to identify the gene knockout targets of recombinant
E. coli
by resorting to various optimization techniques, for example,
iterative linear programming, mixed-integer optimization [82], quad-
ratic programming [83], and bilevel optimization [84]. As a whole, in
silico and wet experiments are interactively conducted under the
perturbing conditions of biotechnological relevance. The results are
compared for the generation of new knowledge, which is used to refine
the model and to design new experiments. This process is iterated
along systems biotechnological research cycles until the development
of an improved microorganism having desired traits is accomplished.
As shown in figure 7.9, there exist two adaptive feedback loops
and one interactive cycle. The in silico model and experimental design
are iteratively modified through respective feedback loops [14,18,85].
This conventional iterative model-building process, however, may
be insufficient to validate the hypothetical model and experimental
results. Therefore, another interactive communication cycle between
in silico and wet experiments makes such a process more efficient
[17,79,86]. Computational models can be directly refined from new
experimental data, and subsequently the resultant in silico perturba-
tions of the refined model can suggest new experimental design. Thus,
new knowledge can be effectively generated after several iterations
and interactions, thereby facilitating the improvement of strains suitable
for successful industrial applications. Consequently, the essence of
systems biotechnology resides in the integration of wet and dry exper-
iments to achieve a goal of rational metabolic design. In the following,
two examples are presented to illustrate the truly enormous range
of possibilities for the systems biotechnological strategy afforded by
high-throughput experiments and in silico modeling and simulation
(case studies 3 and 4).
Search WWH ::
Custom Search