Fig. 5

Cell type-specific characterization of transcriptomic differences between organoids from DPD deficiency affected and non-affected subjects using BEDwARS deconvolution of bulk RNA-seq data. A UMAP plot of processed cells clustered into 17 groups. Cells on the right represent the affected patient and cells on the left represented the non-affected subject. B Average expression of 11 marker genes in cells of each cluster indexed by numbers. These markers were used for cell type assignment. C Comparison between the average inferred proportions (plain) from eight bulk samples of an affected and a non-affected subject and the average of “true” proportions (diagonal striped) derived from the semi-matched single cell data on three organoids for the same subject. D Inferred proportions of different cell types obtained by BEDwARS deconvolution of bulk RNA-seq data from organoids derived from two non-affected subjects and three affected patients. CN-NEU sum of inferred proportions of CN and NEU, CBC-INTER sum of inferred proportions of CBC and INTER. E Negative logarithm (base 10) of the p-value (\(-{\mathrm{log}}_{10}(\mathrm{pvalue})\)) of hypergeometric tests of Gene Ontology (GO) term enrichments in the top 200 differentially expressed genes (DEGs) from bulk samples (72 affected vs 48 non-affected, “Bulk”), bootstrapped bulk profiles derived from the single cell data (100 affected vs 100 non-affected, “BS-Bulk”) and the cell type-specific bulk expression derived from BEDwARS deconvolution (72 affected vs 48 non-affected, “CN-NEU”, “AS”, “CBC-INTER”, “PGC”, “BRC”, “NEC”). None of the GO terms enriched in the top 200 DEGs of cell type-specific profiles are enriched in the top 200 DEGs derived from the bulk expression profiles. See Additional file 1: Table S3 for information on GO terms. F Logarithm (base 2) of the fold-change (\({\mathrm{log}}_{2}\mathrm{FC}\)) of expression of the 139 genes grouped into cluster 8 based on their pattern of differential expression in different cell types. G The 139 genes of cluster 8 are highly enriched in the GO term “cytosolic ribosome” (hypergeometric test p-value \(2\times {10}^{-85}\))