Fig 1.
Schematic illustration of SCOM.
(a) An illustration of the ncRNA synergistic competition hypothesis. ncRNA A and ncRNA B acting as ceRNAs collectively compete with mRNAs for soaking up miRNAs. In the context of ceRNA competition, the crosstalk between ncRNA A and ncRNA B is in the form of synergistic competition. (b) For each tumor type in Pan-cancer, SCOM firstly extends the SPONGE [27] method to predict ncRNA-related (including lncRNA- and pseudogene-related) ceRNA network by incorporating gene (including miRNAs, lncRNAs, pseudogenes, and mRNAs) expression data in matched tumor tissues and priori information of miRNA-target (including miRNA-lncRNA, miRNA-pseudogene and miRNA-mRNA) interactions. By integrating gene (including lncRNAs and pseudogenes acting as ceRNAs, and target mRNAs) expression data in matched tumor tissues and predicted ncRNA-related ceRNA networks, SCOM utilise three criteria to infer ncRNA synergistic competition network for each tumor type in Pan-cancer. Based on the predicted ncRNA synergistic competition networks across 31 malignant tumors, SCOM further uncovers potential carcinogenic biomarkers.
Fig 2.
The ceRNA regulation landscape across malignant tumors.
(a) The number of ncRNA-related ceRNA interactions in each malignant tumor. (b) The topological characteristics of ncRNA-related ceRNA networks for the 31 malignant tumors. The Degree column is the p-value of fitting power-law distribution. The p-values equal to or more than 0.05 indicate that the degree of ncRNA-related ceRNA networks obeys the power-law distribution. The Path and Density columns represent the–log10(p-value) of Student’s t-test, and the p-values less than 0.05 display that the characteristic path lengths (or densities) of ncRNA-related ceRNA networks are significantly shorter (or higher) than those of their corresponding random networks. (c) The similarity matrix shows the similarity between each pair of lncRNA (upper triangle) and pseudogene (lower triangle) related ceRNA networks across 31 malignant tumors. (d) The radar chart shows the percentage of ceRNA regulations predicted in different numbers of malignant tumors. (e) The conserved ceRNA regulation predicted in 31 malignant tumors.
Fig 3.
Network analysis of ncRNA synergistic competition across malignant tumors.
(a) The number of ncRNA synergistic competition interactions in each malignant tumor. (b) The topological characteristics of ncRNA synergistic competition networks for the 31 malignant tumors. The Degree column is the p-value of fitting power-law distribution. The p-values equal to or more than 0.05 indicate that the degree of the ncRNA synergistic competition networks obeys the power-law distribution. The Path and Density columns represent the–log10(p-value) of Student’s t-test, and the p-values less than 0.05 suggest that the characteristic path lengths (or densities) of ncRNA synergistic competition networks are significantly shorter (or higher) than those of their corresponding random networks. (c) The similarity matrix showing the similarity between each pair of ncRNA synergistic competition networks across 31 malignant tumors. (d) The radar chart showing the percentages of ncRNA synergistic competition interactions predicted in different number of malignant tumors. (e) The bar chart showing the percentages of ncRNA synergistic competition interactions predicted in one malignant tumor.
Fig 4.
Hub analysis of ncRNA synergistic competition across malignant tumors.
(a) The number of hub ncRNAs in each malignant tumor. (b) The radar chart displaying the percentage of hub ncRNAs predicted in different number of malignant tumors. (c) Multi-class classification analysis of conserved hub ncRNAs in at least 16 malignant tumors. (d) Survival analysis of conserved hub ncRNAs in at least 16 malignant tumors. The symbol “/” stands for an infinite value. (e-g) Enrichment analysis (three categories including disease, cell marker and cancer hallmark) of conserved hub ncRNAs in at least 16 malignant tumors. (h) The bar chart indicating the percentage of hub ncRNAs predicted in one malignant tumor.
Fig 5.
Drug resistance analysis of ncRNA synergistic competition networks and hub ncRNAs across malignant tumors.
(a) The number of significantly enriched drugs for each malignant tumor. (b) Relationships between ncRNA synergistic competition networks and enriched drugs. For example, the node “Network_ACC” denotes the ncRNA synergistic competition network in ACC. Larger drug nodes denote more links with ncRNA synergistic competition networks in malignant cancers. (c) Relationships between hub ncRNAs and enriched drugs. For example, the node “Hub_ACC” denotes the hub ncRNAs in ACC. Larger drug nodes denote more links with hub ncRNAs in malignant cancers.
Fig 6.
Molecular subtype analysis of malignant tumors.
(a) Molecular subtyping of the 31 malignant tumors using ncRNA synergistic competition networks and hub ncRNAs. Higher average silhouette width indicates that a malignant tumor sample is better matched with the identified molecular subtypes. (b) Molecular subtyping of LGG using ncRNA synergistic competition network. (c) Molecular subtyping of LGG using hub ncRNAs. (d) Classification analysis of BRCA subtypes.
Fig 7.
Immune regulation analysis of synergistic competition ncRNAs across malignant tumors.
(a) The number of significant immune-related pathways associated with each malignant tumor. (b) The number of significant immune cells associated with each malignant tumor. (c) Relationships between ncRNA synergistic competition networks and immune-related pathways. For example, the node “Network_ACC” denotes the ncRNA synergistic competition network in ACC. Larger immune-related pathway nodes denote more links with ncRNA synergistic competition networks in malignant cancers. (d) Relationships between hub ncRNAs and immune-related pathways. For example, the node “Hub_ACC” denotes the hub ncRNAs in ACC. Larger immune-related pathway nodes denote more links with hub ncRNAs in malignant cancers.
Fig 8.
(a) The search of ceRNA regulation in Pan-cancer. (b) The search of ncRNA synergistic competition in Pan-cancer. (c) The search of hub ncRNAs in Pan-cancer. (d) All resources are provided to be downloaded for further analysis. (e) The SCOM method is available for users to study ncRNA synergistic competition by using their own datasets.