Fig. 1

Integrative co-expression analysis can resolve constituent cell type identities from unfractionated human stomach tissue RNAseq data. (A) RNAseq data for 359 unfractionated human stomach samples were retrieved from GTEx V8. Each sample contained (i) mixed cell types, which contributed (ii) differing proportions of sequenced mRNA. (B) To profile cell type-enriched transcriptomes, constituent cell types were identified and candidate marker genes (´reference transcripts´ [Ref.T.]) for virtual tagging of each were selected, based on in house tissue protein profiling and/or existing literature and datasets. (C) Matrix of correlation coefficients between selected Ref.T. across the sample set. (D) Mean correlation coefficients of genes above designated thresholds for classification as cell-type enriched in stomach: (i) parietal cells [PC], (ii) chief cells [CC], (iii) gastric enteroendocrine cells [GEEC], (iv) gastric mucous cells [GMC], (v) mitotic cells [MTC], (vi) endothelial cells [EC], (vii) fibroblasts [FB], (viii) macrophages [MC], (ix) neutrophils [NP], (x) T-cells [TC], (xi) plasma cells [PC] with all Ref.T. panels. (E) Over-represented gene ontology terms among genes predicted to be: (i) endothelial cell, (ii) fibroblast or (iii) T-cell enriched. (F) Principal component analysis of correlation profiles of cell type enriched genes. See also Table S1 Tab 1 and 2 and Figure S1 for method overview