Fig. 2

Functional analysis of the predicted candidates. a Functional similarity between the six core member genes and the predicted candidates. For each core member gene, only those genes having nonzero similarity to the corresponding core member gene were used in GSEA. We ranked these genes (x-axis) based on their functional similarity to the corresponding core member gene to calculate the enrichment score (ES, y-axis) by hitting the tested gene set-the predicted candidates. The number of hits in the candidate gene set is denoted as 'hit,' while 'miss' represents the number of genes without hits. The p-values (P) and z-scores were derived from 10,000 random permutations of gene ranking during the GSEA process. The maximal enrichment scores (ES) are marked as red circles. b Highly-voted functional modules within the candidates for each cancer type. The details of identifying the highly-voted functional modules formed by the predicted candidates in each cancer type are described in the Supplementary Information. Each pie chart shows the functional modules in which the highly voted candidates are involved in the corresponding cancer type. The percentage denotes the relative significance of one functional module to the others. Color coding is used to highlight functional modules associated with notable biological processes: red (positive regulation of tumor necrosis factor production or cellular response to cytokine stimulus), blue (immune response), green (negative regulation of cell death), and orange (cancer metastasis). The biological processes with long descriptions are represented by their GO terms (Additional file 1: Table S2)