In Silico Analysis of Long–Run Tolerance of Metabolic Networks to Transient Random Faults

In this paper we aim to describe the propagation of a ”failure event” through a metabolic network (i.e. no formation of a set of metabolites) and, by means of a Markov chain, to be able to compute the long–run failure probability of each metabolite following a transient random failure event occurring somewhere in the network. Moreover, we will introduce and motivate the definition of a ”tolerance index” for each metabolite. The in silico analysis reveals that in most of the 142 organisms’ metabolic network we considered, such tolerance index is approximately uniformly distributed among nodes with the same in–degree, while the failure distribution and the connection degree is not. A list of metabolites ranked by failure probability and tolerance index for all the 142 organism is also provided in the supplementary material to this paper (http://www.dis.uniroma1.it/ ̃farina/robustness/).