Fig. 2: Identification of the cancer stem cells (CSC) signature using weighted gene co-expression network analysis (WGCNA).

a UMAP plots showing the grouping of malignant cells (left) and non-tumor cells (right) based on sample origins, respectively. b Partitioning of 25,839 malignant cells into 320 metacells. Metacell clusters with the same color had relatively similar expression patterns. c Construction of a scale-free topology model with a soft thresholding power set to 9. d Hierarchical cluster tree revealing a total of 27 modules identified by WGCNA. e The upper dendrogram plot exhibits the hierarchical clustering of modules obtained from the clustering analysis depicted in Fig. 2d. The branches in the dendrogram represent the grouping of modules that display positive correlation. Meanwhile, the lower heatmap plot illustrates the adjacencies among all modules. Each row and column in the heatmap corresponds to a specific module, denoted by color. In the heatmap representation, the color scheme denotes the degree of adjacency between modules. The blue color signifies low adjacency, indicating a negative correlation, while red denotes high adjacency, indicating a positive correlation. f Heatmap displaying the correlation of each module with the CytoTRACE score, with the first number in each grid denoting the correlation coefficient and the second reflecting the p-value. g Identification of 107 CSC-related genes as the signature of CSC by intersecting the marker genes of malignant cells with the genes in the MEturquoise module.