Fig. 4

Biological and clinical significance of the BCSC highly expressed genes. a Gene Ontology (GO) analysis of the BCSC highly expressed genes in biological process. P values (one-tail Fisher exact P values used for gene enrichment analysis) were calculated in the DAVID database (https://david.ncifcrf.gov/tools.jsp). b Interaction network of BCSC highly expressed genes integrated from the STRING database. Network nodes represent genes, and edges represent gene–gene associations. A detailed legend is available at https://string-db.org. c Investigation of the clinical relevance of BCSC highly expressed genes in 22 cancer types. The expression of each gene in cancer and corresponding normal tissues was analyzed by a two-tailed t test. Heatmap was horizontally sorted by the number of genes with P < 0.01 in a particular cancer type, shown as red columns on the top. d Kaplan–Meier relapse-free survival curve (left) of patients with low (green) and high (red) risk grouped by BCSC highly expressed genes in SurvExpress (dataset: Breast cancer relapse data). The total number of each group was shown in the top right corner, and the number of censoring samples are marked with a “+” symbol. The concordance index (CI) per curve was also included. The P value was determined by a log-rank test. The x axis represents the years of the study. In rows and corresponding colors, the numbers of samples not presenting the event at the matching time are shown. The box plot (right) shows the comparison of the gene expression between the low- and high-risk groups. Genes significantly (P < 0.05) highly expressed in the high-risk group are highlighted in red. P values were calculated using two-tailed t test