Skip to main content
Fig. 8 | Genome Medicine

Fig. 8

From: A reference profile-free deconvolution method to infer cancer cell-intrinsic subtypes and tumor-type-specific stromal profiles

Fig. 8

Comparison of scRNAseq data and DeClust tumor-type-specific stromal profiles. Clustering results of scRNAseq data for ccRCC samples (a) and BLCA samples (d). Cells are colored according to cell clusters. Expression levels (heat map) of cell type-specific markers (top) and EMT genes (bottom) in each cell of the RNAseq data for ccRCC (b) or BLCA (e). Marker gene expression levels are represented as the absolute expression value, log2(read count+‚ÄČ1), while expression levels for EMT genes are represented as the row-wise scaled relative expression value. For display purpose, cells in ccRCC2 were downsampled to the same size of ccRCC1. Cells in Epithelia-c0 and Epitheila-c1 were downsampled so that the combined cell number in the two clusters equals to that in Epithelial-c2. Correlation between stromal profiles estimated by DeClust and stromal profiles calculated by mean expression profile of each stromal cell cluster in scRNAseq data of ccRCC (c) or BLCA (f). Error bars indicate the 95% confidence interval

Back to article page