Skip to main content
Fig. 1 | Genome Biology

Fig. 1

From: A human lung tumor microenvironment interactome identifies clinically relevant cell-type cross-talk

Fig. 1

a Schema for dissociation, flow-sorting, and RNA-seq profiling. b Multidimensional scaling analysis of transcriptomes of cell types sorted from surgically resected primary human NSCLC tumors. Axis units are arbitrary. Cell types are depicted by colors as in 1a. c Unbiased hierarchical clustering of sorted samples. d Top 25 most differentially expressed genes between malignant cells from adenocarcinoma and SCC. e Comparison of bulk vs reconstituted transcriptomic profiles. Shown are average values across all samples for each gene measured by RNA-seq. Panel below shows functional enrichment of genes higher in bulk for tissues that were not sorted for profiling. f Average percentage difference in immune cell types deconvolved in bulk vs sorted CD45+ populations showing enrichment of activated mast cell profiles by sorting, and conversely loss of plasma cells. g CIBERSORT deconvolution of immune populations in adenocarcinoma (pink) and SCC (light blue) identifies similarities and differences in immune cell types that are relatively depleted (below diagonal) or enriched (above diagonal) by dissociation and sorting. MC+ = activated mast cells; PC = plasma cells; M2 = M2-polarized macrophages; MemB = memory B cells; CD8 = CD8 T cells; Eos = eosinophils

Back to article page