Design of Pathway-Level Bioprocess Monitoring and Control Strategies Supported by Metabolic Networks - Measurement, Monitoring, Modelling, and Control of Bioprocesses - page 203

Moreover, metabolites are not organism specific, which means the techniques are
equally applicable to prokaryotic, fungal, plant and animal cells.
Even though a lot of progress has been made towards enabling whole metab-
olome quantification, these techniques still face challenges related to the inherent
characteristics of the metabolome. The size of the metabolome varies greatly,
depending on the organism studied. The nature of the metabolites, whether they
are polar or nonpolar, volatile or nonvolatile, also influences the analysis, and most
methods are biased towards some group of metabolites. In addition, the concen-
tration of different metabolites extends over several orders of magnitude [ 48 ], thus
adding difficulty to the task of quantifying all metabolites with a single technique.
However, quantification of the whole metabolome is not essential for the purposes
of process monitoring and control, as a subset of key metabolites is enough to infer
cell function.
4 Elementary Mode Reduction
The number of elementary modes increases geometrically with the size of the
network. The typically very high number of elementary modes denotes the innate
adaptability and robustness of biological networks. As a consequence, the com-
putation of elementary modes suffers from combinatorial explosion, particular for
genome-scale networks. The central carbon metabolism of a genome-scale
reconstructed E. coli metabolic network has approximately 26 million EMs [ 49 ]. It
is essential to reduce such large numbers of elementary modes according to some
criterion in order to decrease the computational power requirements. Indeed, not
all calculated elementary modes are thermodynamically feasible or even physio-
logically reachable [ 50 ]. Several methods have been developed to reduce the
number of elementary modes, founded on different principles. In what follows we
review some of them.
4.1 Reduction Based on Network Structural Properties
Elementary modes can be reduced on the basis of structural information of the
metabolic network without the use of experimental data. de Figueiredo et al. [ 51 ]
presented a method based on the ranking of elementary modes in increasing order
of number of reactions. This approach enables identification of the K shortest
elementary modes, which are in principle energetically more efficient. Song and
Ramkrishna [ 14 ] proposed a reduction algorithm based on the effect of elementary
modes on the convex hull volume. The principle consists in removing the ele-
mentary modes with negligible contribution to the convex hull volume of the
original network. This allowed a priori reduction from the initial 369 to a final set