Fig. 7
From: Unsupervised deep learning reveals prognostically relevant subtypes of glioblastoma
Word clouds for the top 10 DEGs or mutations with the largest correlations with each GBM cluster. Each GBM cluster is represented by a word cloud of genes, colored according to the corresponding curve in the Kaplan-Meier plot (Fig. 5b). The size and color of each gene in each word cloud correspond to the relative strength of the correlation for that gene. The largest and darkest words in the word clouds correspond to the strongest correlations. Each cluster’s word cloud was created independently. Therefore, the differential sizes of each gene (representing correlations) is only relevant when compared to other genes in that cluster. Gene sizes should not be compared across different clusters