Fig. 3
From: GSAE: an autoencoder with embedded gene-set nodes for genomics functional characterization
Subtype analysis in BRCA data set. (a) The t-SNE results of BRCA data, where HDBSCAN classified the samples into two groups. The noisy samples were labeled in black and omitted from further analysis. (b) The density plots of the most significant up-superset and three selected top gene sets. The blue/yellow arrow corresponds to positive/negative weight in the model between the gene set and superset. (c) The density plots of the most significant down-superset and three selected top gene sets. (d) The Venn diagram of the significant gene sets in the top 3 up-supersets