Genome-Scale Models of Metabolic and Regulatory Networks - Systems Biology: Networks, Models, and Applications - page 232

network as a basis for simulations as described later in this chapter.
For example, the known physiology of an organism may indicate
that it is capable of producing a particular amino acid, but steps in the
biosynthetic pathway for this amino acid are missing. In this case,
reference pathways from other organisms can be used to establish
potential missing reactions and directed homology searches can be
used to find gene associations for these reactions [6,14]. Similarly, often
the transport mechanisms that are used to transport metabolites in
and out of the cell are not known, and, based on known physiology,
the ability to transport metabolites may have to be assumed without
known transport facilitators. This type of careful analysis relies largely
on topics and reviews written on the genetics, biochemistry, molecular
biology, and physiology of particular organisms such as E. coli or yeast
[15,16].
In order to enable mathematical analysis of network function,
the reactions in the network also need to be mass balanced [14].
This requires estimating the ionization states of metabolites at a par-
ticular pH and including protons and water molecules as part of
the reaction formulas. A further complication that arises in metabolic
network reconstruction for eukaryotic organisms is assigning the
correct intracellular compartments for metabolites and reactions in
the network. This type of information is not typically available in
metabolic network databases and has to be collected either from stud-
ies of individual gene product localizations published in the literature
or from high-throughput protein localization screen data [17,18].
A genome-scale metabolic network reconstruction is basically defined
by the components listed above: the correctly balanced stoichiometry
of each individual metabolic reaction including transport reactions,
and the subcellular localization assignment of each metabolic reaction
in the network. While this information already allows analysis of
network structure and capabilities, further details are needed to be
defined in order to connect reactions in the network to the genes and
proteins in an organism. These connections are established through
gene-protein-reaction associations [14,19] that describe, for example,
how two alternative isozymes can catalyze the same reaction or
how two components of a complex come together to catalyze one
reaction.
Genome-scale metabolic network reconstructions have been developed
for a number of organisms, including H. influenzae [20], H. pylori [21],
S. aureus [22], E . coli [14], S. cerevisiae [19,23], S. coelicolor [125], and the
human mitochondrion [24]. Table 8.1 summarizes the properties of these
reconstructions including the numbers of genes, metabolites, and reac-
tions. In the following we will focus primarily on the work done with
the E. coli [14,25] and S. cerevisiae metabolic network models [19,23],
since these models have been most extensively tested experimentally.