Biology Reference
In-Depth Information
is known to give metabolic networks their tremendous versatility and robustness
in the presence of exogenous disturbances [62]. The issue of robustness on scale-
free network in random attacks has received significant attention [74] given its
implications for various real-life networks that possess that property. A measure
for deciding the stability of the network in the presence of random attacks, that is
failures of nodes at random, is quantified by estimating the average diameter of
the network is often defined as the mean shortest path of the network once nodes
are eliminated.
Fig. 3.10. Power-law distribution of network connectivity for experiment-specific gene interaction
network based on the burn-induced inflammation data.
In order to assess the intrinsic characteristic of the protein interaction networks
(Fig. 3.9) determined based on the informative genes that were identified as most
responsive to the inflammatory response, we performed simulations to assess the
stability of the network during failures (elimination of random nodes) and targeted
attacked (elimination of nodes based on their connectivity). It has already been
demonstrated with large networks exhibiting both scale-free and random (expo-
nential) [44] the fundamental differences of the networks. Specifically, random
networks have similar responses, in term of changes in network diameter, in both
random failures and attacks due to the fact that no nodes exhibit higher connec-
tivity and therefore the contributions of all nodes are equivalent. We use network
diameter as a surrogate of a metric for quantifying effects on network structure.
More appropriate metrics could have been adopted as well [76]. Random net-
works (exponential) are expected to show a similar increase in network diameter
under, both, failure and attacks. However, scale-free networks are far more robust
to random failures because the network can always reroute the flow of information
Search WWH ::
Custom Search