Defining Diffuse Large B-Cell Lymphoma Immunotypes by CD8+ T Cells and Natural Killer Cells

Figure 4

The top 12 most important genes were selected for the construction of machine learning models on the training sets. The top 10 most important genes were selected by a random forest model on the expression data from GSE11318 (a), GSE21846 (b), GSE23501 (c), GSE32918 (d), and TCGA-DLBCL (e). The top 6 upregulated and top 6 downregulated genes in the inflamed subtype (IS) with the highest mean importance value were selected for model construction (f).

(a)

(b)

(c)

(d)

(e)

(f)