Genome-Scale Models of Metabolic and Regulatory Networks - Systems Biology: Networks, Models, and Applications - page 231

and analysis of transcriptional regulatory networks is reviewed. The
focus of this section will be on illuminating the differences between the
reconstruction and analysis of metabolic and regulatory networks [9].
We will conclude with discussion of methods that have been developed
to analyze the integrated function of metabolic and regulatory networks
at the genome scale.
METABOLIC NETWORKS
Reconstruction of Metabolic Networks
The publication of the first prokaryotic genomic sequence [2] signaled
the beginning of a new era, not only in experimental biology, but
also in modeling and analysis biochemical reaction networks. For
the first time it was possible to define exactly the genetic makeup of
an organism and hence, in principle, all the biochemical reactions that
can take place in a particular cell. Metabolic physiology has been exten-
sively studied and the reaction stoichiometry of most commonly
occurring metabolic reactions has been well established [10]. All these
factors together allow using comparative genomics to define potential
enzymatic functions for genes and build pathways and networks by
connecting individual enzymatic conversions [11]. In order to assign
enzymatic functions by comparative approaches, a similar function has
to have been characterized for a gene product in another organism.
In the case of prokaryotes, the most common model organism for
metabolic physiology is the gram-negative bacterium Escherichia coli
whereas for eukaryotes the equivalent position is held by the yeast
Saccharomyces cerevisiae .
Metabolic network reconstruction benefits enormously from the avail-
ability of well-curated and comprehensive databases such as KEGG
[12] and EcoCyc [13] that store information about individual metabolic
reactions and whole metabolic pathways. The information in these
databases is usually derived from both manual curation based on
primary literature and automated sequence-based annotation. These
databases form the basis for reconstructing more detailed models
of metabolic networks, by providing comprehensive summaries of the
known and hypothetical metabolic reactions present in an organism.
However, a network reconstructed on the basis of database informa-
tion alone would not function as a predictive model for a number of
reasons.
Despite the best efforts of database curators, many of the networks
contain gaps due to lack of direct biochemical information for a par-
ticular organism and extra reactions due to erroneous sequence
homology-based function assignments. These types of errors can often
be partially fixed by analysis of the network structure and by using the