We address the problem of finding large-scale functional and structural relationships between genes, given a time series of gene expression data, namely mRNA concentration values measured from genetically engineered rat fibroblasts cell lines responding to conditional cMyc protooncogene activation. We show how it is possible to retrieve suitable information about molecular mechanisms governing the cell response to conditional perturbations. This task is complex because typical highthroughput genomics experiments are performed with high number of probesets (10 3 − 10 4 genes) and a limited number of observations (< 10 2 time points). In this paper we develop a deepest analysis of our previous work [Remondini et al., 2005] in which we characterized some of the main features of a gene-gene interaction network reconstructed from temporal correlation of gene expression time series. One first advancement is based on the comparison of the reconstructed network with networks obtained from randomly generated data, in order to characterize which features retrieve real biological information, and which are instead due to the characteristics of the network reconstruction method. The second and perhaps more relevant advancement is the characterization of the global change in co-expression pattern following cMyc activation as compared to a basal unperturbed state. We propose an analogy with a physical system