Fig. 5

The TCGA-110-CL: an improved cell line panel integrating TCGA and CCLE data. a Heatmap of correlations between cell lines in the NCI-60 panel and primary tumor data. Only 36 cell lines which are shared between NCI-60 panel and CCLE are shown. The tumor types of each cell line are indicated by the annotation bar to the left of the heatmap. b Heatmap of improved NCI-60 panel. Improved panel has the same number of cell lines and tumor types as the original NCI-60 panel, but the cell lines with the highest correlations with their matched primary tumor samples were selected. c Boxplot showing that the improved NCI-60 panel has significantly higher correlations (two-sided Wilcoxon test p-value = 7.6e-07) with their matched primary tumor samples compared with the original NCI-60 panel. The center line in the boxplot depicts the median, the box limits depict the upper and lower quartiles, and the whiskers depict 1.5 times the interquartile range (IQR). d Proposed TCGA-110-CL panel. An improved cell line panel that includes five cell lines with the highest correlations to their matched primary tumor samples across 22 tumor types. For the tumor types with subtype predictions, cell lines with the highest correlations within each subtype were chosen to maximize subtype representation in the panel