Fig 1.
Expression-based unsupervised hierarchical clustering.
(A) The expression data of 206 digestive cancer cell lines from the Cancer Cell Line Encyclopedia (CCLE) were analyzed. (B) The 28 liver cancer cell lines obtained from CCLE were extracted and analyzed again. (C) Nineteen liver cancer cell lines in another dataset from the Catalogue of Somatic Mutations in Cancer (COSMIC) were clustered.
Table 1.
Details of liver cancer cell lines.
Fig 2.
Heatmap visualization and hierarchical clustering of the selected differentially expressed genes.
Heatmap showing the hierarchical clustering of (A) 22 liver cancer cell lines from the Cancer Cell Line Encyclopedia (CCLE) and (B) 371 hepatocellular carcinoma (HCC) tumor samples from the Cancer Genome Atlas (TCGA) based on their gene expression. Each row represents one sample (cell line or HCC), and each column represents the expression level of one gene.
Table 2.
Clinical information of HCC of TCGA.
Fig 3.
Gene Ontology (GO) analysis of A and B genes using DAVID.
The enriched terms in (A) the A genes are indicated by a reddish column, and those in (B) the B genes are indicated by a bluish column. The color depth corresponds to the number of counts. Black vertical lines show the -log(0.05) results for (A) A genes and (B) B genes.
Fig 4.
Unsupervised hierarchical clustering of liver cancer cell lines and primary cells.
The cluster, including cell line A and hepatocytes, is indicated with a red line on the right, and the cluster, including cell line B and stromal cells, is indicated with a blue line.
Fig 5.
Comparison of the vector values of the expression pattern for each cell type.
The graph of vector values shows primary cells that resemble cluster B cell lines. The color of columns for each cell corresponds to the cell category, same as in Fig 4.
Fig 6.
Unsupervised clustering analysis of cell lines established from bone, breast, or skin cancer.
A cluster of fibroblast-like cell lines was observed regardless of the origin. In correlation analysis with the representative vector, setting the value of primary fibroblasts to 1, a fibroblast-like cell line has a value >0.88. Those with a value of 0.88 or less were classified as NON.
Table 3.
Correlation with the representative vector of fibroblasts.