Introduction et al Nowell, 1976 Genetic mutations of the signaling proteins might overactivate key cell-signaling properties such as cell proliferation or survival and then give rise to the cell with selective advantages for uncontrolled cellular growth and promoting tumor progression. In addition, mutations may also inhibit the function of tumor-suppressor proteins, resulting in a relief from normal constraints on growth. Furthermore, epigenetic alterations by promoter methylation, resulting in transcriptional repression of genes controlling tumor malignancy, is another important mechanism for the loss of gene function that can provide a selective advantage to tumor cells. et al et al et al et al http://www.sanger.ac.uk/genetics/CGP/ et al et al et al et al Results and discussion et al et al et al et al et al et al Wang and Purisima, 2005 et al et al et al et al et al et al et al et al Figure 1A Supplementary Figure 1 Supplementary Figure 1 et al et al Cancer mutated genes are enriched in signaling hubs but not in neutral hubs P −6 Supplementary Figure 2 P −14 Supplementary Figure 2 R P −16 R P et al P P P Supplementary Figure 2 Activating and inhibitory signals enhance and alleviate oncogene-signaling flows, respectively Figure 1B Figure 1B Figure 1B Figure 1B Figure 1B Table I Table I Figure 1B Table I P et al et al Mutated and methylated genes are enriched in positive and negative regulatory loops, respectively et al Wang and Purisima, 2005 Supplementary Figure 3 P −9 Supplementary Figure 3 Supplementary Table 1 Figure 2 Supplementary Table 1 Figure 2 Supplementary Table 2 Supplementary Figure 4a–d et al Ferrell, 2002 et al et al et al et al An oncogene-signaling map emerges from the network P −4 Figure 3 P −16 P Supplementary Table 3 Supplementary Table 4 As mentioned above, oncogene-signaling-dependent events, which we define as the interactions between the cancer mutated or methylated genes, are frequently found in tumor samples and represent various oncogene-driving events that could play more critical roles in generating tumor phenotypes. To systematically identify such events and discover how they are organized in the map, we charted the gene mutation frequency onto the map and highlighted the signaling links between any two genes that have high mutation frequencies. Most genes have mutation frequencies lower than 2%, whereas a handful of genes have very high mutation frequencies, such as p53 (41%), PI3K (10%) and RAS (15%) (see Materials and methods). Therefore, a gene mutation frequency equal to or greater than 2% was considered as high. Interestingly, nearly 10% of the links in the map are oncogene-signaling-dependent events. Certain signaling events such as Pten-PI3K and RAS-PI3K in the map are well-known oncogene-signaling-dependent events/cascades that are frequently used in various cancers. Figure 3 Hanahan and Weinberg, 2000 Siegel and Massague, 2003 Functional collaboration of genes between oncogene-signaling blocks et al Supplementary Table 5 Supplementary Table 5 http://david.abcc.ncifcrf.gov/home.jsp Supplementary Table 6 Supplementary Table 7 M s,b M s,b Figure 4A Supplementary Table 7 Figure 4A P −4 Supplementary Table 5 Figure 3 P −4 P P P −4 Meuwissen and Berns, 2005 et al Supplementary Figure 5 et al et al Figure 4B Figure 4B P −4 et al Figure 4C and D Figure 4A Figure 5 Figure 5 et al Figure 1A Supplementary Figure 6a and b The mutated genes in the network provide a predictive power Supplementary Table 8 Figure 6 et al Concluding remarks Although a wide variety of genetic and epigenetic events contribute to the signaling of tumorigenesis, it has been challenging to gain a global view of where and how they affect the signaling alterations to generate tumors on the entire signaling network. By integrative analysis of the human signaling network with cancer-associated mutated and methylated genes, we uncovered an overall picture of the network architecture where the oncogenic stimuli occur and the regulatory mechanisms involving mutated and methylated genes. Mutations, the majority of which are activating, preferentially occur in the signaling hub genes (but not neutral hubs) and the genes of the positive regulatory loops, whereas methylated alterations tend to occur in the genes of the negative regulatory loops. Cancer and cell signaling have been well established, and extensive efforts have been made to illustrate cancer signaling during the past few decades. However, it has been a struggle to get clues of how the oncogene signaling is structurally and functionally organized. In this analysis, we extracted an oncogene-signaling map, which provides a blueprint of the oncogene signaling in cancer cells. From the map, we discerned that the oncogene-signaling-dependent events form three highly connected regions that resemble oncogene-signaling superhighways frequently used in tumorigenesis. Topologically, the map has been divided into 12 oncogene-signaling blocks. Functional collaborations between subsets of these blocks are underlying tumorigenesis. In most tumors, genes in both p53 and RAS blocks often get mutated, although the combinations of p53 with other signaling blocks are also found in a small fraction of tumors. Analysis of the NCI-60 cell line panel mutations showed the enrichment of gene mutations in p53 and RAS blocks, which is similar to the patterns found in the 592 samples. Furthermore, we can dissect some of this functional collaboration among different tumor types. These facts indicate that at least two signaling gene mutations, one from the p53 block and the other from another block, are necessary for tumorigenesis. This fact supports the notion that both the prevention of cell death (p53 block) and the promotion of cell proliferation (RAS or other blocks) are necessary to generate most tumors. Chng, 2007 Kaiser, 2006 et al et al Chanock and Thomas, 2007 et al et al in toto Liu and Lemberger, 2007 Materials and methods Data sets used in this study Human signaling network http://www.biocarta.com/ et al et al http://cancer.cellmap.org/cellmap/ Supplementary Table 9 Cancer mutated genes http://www.sanger.ac.uk/genetics/CGP/cosmic/ et al et al et al et al et al et al et al et al et al et al Supplementary Table 10 Supplementary Table 10 et al et al et al et al Figure 1A Supplementary Table 10 Methylated genes in cancer stem cells et al et al et al Supplementary Table 11 Cancer-associated gene set http://plasmid.hms.harvard.edu/GetCollectionList.do Supplementary Table 8 Microarray data et al Oncogenic map extraction Supplementary File 1 Network analysis et al Newman, 2006 Analyzing the enrichment of the mutated and methylated genes in the network motifs We mapped the mutated and methylated genes onto each type of the motifs. We then counted the number of mutated or methylated genes in each motif and classified each type of motif into several subgroups based on the number of nodes that are mutated or methylated genes. We then calculated the ratio (Ra) of the activation links to the total activation and inhibitory links in each subgroup. Randomization tests Wang and Purisima (2005) Survival analysis http://www.r-project.org/ P Supplementary Material Supplementary Figures  Supplementary Table 1  Supplementary Table 2  Supplementary Table 3  Supplementary Table 4  Supplementary Table 5  Supplementary Table 6  Supplementary Table 7  Supplementary Table 8  Supplementary Table 9  Supplementary Table 10  Supplementary Table 11  Supplementary Java File  Supplementary Tables and Java File Legends