Extended Data Fig. 5: BayesPrism expression estimates group malignant cells from the same patient. | Nature Cancer

Extended Data Fig. 5: BayesPrism expression estimates group malignant cells from the same patient.

From: Cell type and gene expression deconvolution with BayesPrism enables Bayesian integrative analysis across bulk and single-cell RNA sequencing in oncology

Extended Data Fig. 5

Heatmaps show the pairwise Pearson correlation matrix between gene expression computed for each pair of simulated pseudo-bulk samples with the fraction of malignant cells greater than 10% (top row), 20% (mid) and 50% (bottom). Simulated samples were constructed by drawing a random proportion of each non-malignant cell type and sampling the malignant cells from one of 27 GBM patients. Vectors used for computing the Pearson correlation are of length equal to the total number of genes used to perform deconvolution, denoting zero centered variance-stabilizing transformed reads. Simulated pseudo-bulk samples are grouped by hierarchical clustering, as shown by the dendrogram. Left column: Correlations using total expression from bulk samples without any correction; Right column: Correlations over the same set of samples and genes as in the left column but using deconvolved expression profiles for malignant cells in each sample.

Back to article page