Using HCC and adjacent normal liver samples we investigated the gene and sCNV changes associated with tumorigenesis by comprehensively discovering the significant Chlorhexidine hydrochloride relationships within and between DNA copy number variation, global gene expression in TU and AN tissue and patient survival. Analysis of these data revealed the appearance of highly significant network changes as shown by gene pairs differentially correlated between AN and TU tissue. Interestingly this process largely consisted of loss of correlation in the TU samples consistent with disruption of normal networks. A subset of the changes observed involved gain of correlation in TU indicating the formation of new networks in some cases. Consistent with the view that loss of connectivity may represent loss of functionality and gain of connectivity may be gain of functionality, the LOC subset of genes was enriched for genes involved in normal liver function that might be expected to be largely extraneous to the needs of the tumor, whereas the GOC subset of genes is enriched in the essential tumor function of cell cycle. This appearance of loss and gain of connectivity during tumorigenesis may therefore be analogous to the long established concepts of tumor suppressers and oncogenes. Given the relative abundance of LOC versus GOC events this implies that tumorigenesis in HCC at least is to a large degree one of disruption of tumor suppressing normal networks. Although smaller in number the GOC genes likely represent functions selected as important for disease progression and as such may be important points of intervention. Genes in TU were found to be strongly associated in cis and in trans with sCNV frequently involving large chromosomal regions. Within the architecture of sCNV-to-gene associations a number of hotspots were found where many more genes were associated with a particular marker than would be expected by chance. Additionally the genes associated with the hotspots were highly overlapping suggesting that multiple different loci may coordinately regulate a core subset of genes. The finding of hotspots in cancer data may not be unique to HCC in that similar associations, even involving the same genes and sCNV loci were found in an independent collection of cancer cell lines. Given the common architecture of sCNV across many tumor types, the cis and trans Lomitapide Mesylate correlations documented here may therefore be relevant to a broad range of diseases. The differentially connected genes between AN and TU tissue were also significantly enriched for association to sCNV markers in TU suggesting that the network transitions and associated functional changes may be mediated by somatic sCNV. A surprising finding in this study was that at the same FDR, three times as many genes predictive of survival were found in AN than in TU tissue. Furthermore, although the AN-survival and TU-survival genes overlapped more than would be expected by chance, the majority of genes in each case were not predictive in the other tissue. A direct connection between the altered predictive value of the genes in AN and TU was found by association to sCNV markers where AN-survival genes were preferentially associated with sCNV markers in TU that were not predictive, and TU-survival genes were enriched for association to predictive sCNV markers. It therefore seems that the sCNV in tumors may be sufficient to explain the transformation of the predictive value of genes in AN versus TU. To directly address the hypothesis of whether the pre-existing state of genes, we measured the transcriptional signature of a treatment that promotes HCC tumorigenesis.