Fig. 5

Residual analysis from high grade serous ovarian cancer (HGSOC) dissociated and classic bulk samples deconvolved with NNLS using matched single-cell RNA-seq data. A–B Comparison of residual values of dissociated bulks (ribosomal RNA depleted (-rRNA)) and classic Bulks (-rRNA) at the gene and sample level. Dissociated bulk residuals are on the x-axis, and classic bulks are on the y-axis in both panels. CIBERSORTx barcode genes (calculated from all single-cell datasets), adipose markers, and dissociation response genes (from the original paper) are compared. A CIBERSORTx barcode genes show no difference between bulks, but adipocyte markers show higher values overall in classic bulks, whereas dissociation response genes have lower values in the residual of dissociated bulks. B Zoomed-in version of panel A along the dotted line, coloring each gene by which sample it originates from. C The difference between in (non-negative matrix factorization) NMF components shows one component, 2, significantly different between the dissociated and classic bulks deconvolved with matched single-cell RNA-seq dataset as reference (left). After deconvolution of the same bulks, but adding an adipose signal to the reference from single-nucleus RNA-seq dataset (right)