Fig. 2From: Comprehensive benchmarking of CITE-seq versus DOGMA-seq single cell multimodal omicsComparison between CITE-seq and DOGMA-seq. A Overview of study design for comparison between CITE-seq and DOGMA-seq. Each of four aliquots of T cells from two human donors were activated and stimulated under a different stimulation condition (total of four stimulation conditions in eight tissue culture wells) in a 12-h tissue culture. The cells from each of the eight tissue culture wells were then labeled with a unique hashtag (total of eight unique hashtags). Approximately equal numbers of the eight uniquely hashtagged cell populations were then pooled and labeled with antibody cocktail. Different aliquots of the hashtag- and antibody cocktail-labeled pool of cells then underwent either CITE-seq or DOGMA-seq with DIG library preparation. B–D Boxplot showing quality control metric comparisons between CITE-seq and DOGMA-seq. Median values are indicated with corresponding colors. B Protein tag complexity per cell. Median ratio = 0.90. C Number of genes per cell. Median ratio = 1.11. D Percentage of UMIs mapped to mtRNA per cell. Median ratio = 0.84. E Pairwise comparison of protein tag detection frequencies in CITE-seq (x-axis) and DOGMA-seq (y-axis). Each point represents a single protein tag. Blue points highlight isotype control protein tags. Grey points are all other protein tags. F (i) Correlation of protein tag fold change (log2) as detected by CITE-seq and DOGMA-seq in two groups of T cells from donor SB775372 that were both activated and cultured with IL-1β and IL-23, and one of the two groups was also cultured with PGE2. Top upregulated protein tags are highlighted in red; downregulated protein tags are highlighted in blue. (ii) in samples from donor SB775393, as (i). G Pairwise comparison of gene detection frequencies in CITE-seq (x-axis) and DOGMA-seq (y-axis). Each point represents a single gene. Blue points highlight exon-dominated genes; red points highlight intron-dominated genes. An exon-dominated gene is defined as a gene with proportion of exonic UMIs (DOGMA-seq) > 0.5. An intron-dominated gene is defined as a gene with proportion of exonic UMIs (DOGMA-seq) ≤ 0.5. NA means proportion of exonic UMIs is not available. H (i) Correlation of gene fold change (log2) as detected by CITE-seq and DOGMA-seq in two groups of T cells from donor SB775372 that were both activated and cultured with IL-1β and IL-23, and one of the two groups was also cultured with PGE2. Selected intron-dominated genes are highlighted in red; selected exon-dominated genes are highlighted in blue. (ii) in samples from donor SB775393, as (i). I (i) “Harmonized” 2WNN UMAP plot showing clusters identified in 3WNN clustering of DOGMA-seq data. (ii) “Harmonized” 2WNN UMAP plot showing cell types predicted after projection into the Azimuth PBMC reference. J (i) “Harmonized” 2WNN UMAP plot showing clusters identified in 2WNN clustering of CITE-seq data. (ii) “Harmonized” 2WNN UMAP plot showing cell types predicted after projection into the Azimuth PBMC reference. K Line plots showing average ROGUE (purity of identified clusters) for clustering based on 3WNN/2WNN, RNA, and ADT spaces, with resolution ranging from 0.02 to 0.3. Number of clusters identified in each clustering was labeled. L Bar plots show that proportions of predicted cell types in DOGMA-seq and CITE-seq data are quite similar. M “Harmonized” 2WNN UMAP plots highlighting canonical markers for Th17 cells in DOGMA-seq data. ATAC marker is motif activity (the deviations in chromatin accessibility across the set of regions related to the motif) calculated from ATAC-seq peaks. Cluster 5 in Fig. 2Ii is enriched for Th17 cells based on CCR4 ADT, CCR6 ADT, and RORC ATAC signals. N “Harmonized” 2WNN UMAP plots highlight canonical markers for Th17 cells in CITE-seq dataBack to article page