Extended Data Fig. 4: Number of cancer-type-specific genes per cell type.
a. Scatter plot to illustrate the definition of cancer-type-specific gene expression at different thresholds. Points correspond to genes whose mean expression in malignant cells is highest in HCC than in all other cancer types. A point’s y axis value denotes the average expression of this gene in malignant cells in HCC, while its x axis value corresponds to the maximum of its mean expression levels in malignant cells across all non-HCC cancer types. Each dashed line denotes a choice of threshold, whereby the number of genes whose expression in malignant cells is specific to HCC is defined as the number of points above this dashed line. b. Line plot showing the median number of cancer-type-specific genes (y axis, median across cancer types) for each cell type (x axis) for different choices of threshold (colour). Cell types are ordered by their average y values. c. Boxplots showing, for each choice of threshold (panels), the log-transformed number of cancer-type-specific genes (y axis) per cancer type (points/colour) for each cell type (x axis). Cell types are ordered as in b. Each point in b corresponds to the median of points for the corresponding box in c, after reversing the log transformation. Boxes indicate the median and 1st and 3rd quartiles, while the upper, respectively the lower whiskers extend to the maximal, resp. minimal values no further than 1.5 times the interquartile range from the 3rd, resp. 1st quartiles. Groups (cell types) consist of n = 20, 18, 21, 15, 17, 10, 11, 22 data points (in order from left to right), corresponding to differences in average expression levels across biologically distinct samples.
