Fig. 4

Hashing of human PBMCs including SARS-CoV-2 clinical samples. Each column represents a separate hashing method or a condition (e.g., SARS-CoV-2, 2nd column). “Pre-sort labeling”—labeling with hashing reagents followed by one wash and live/dead sorting with subsequent loading of the cells on a 10x Genomics chip. Other samples undergone “classical” hashing -labeling followed by 3 washes and loading on a chip. A Hashtag-derived oligo (HTO) matrices were generated using CellRanger, followed by log-transformation and visualised on heatmaps. B MULTISeqDemux-annotated cells (HTO signal) were matched with the freemuxlet-annotated cells (gene data from 3 individuals) and visualized on the gene expression UMAP plots. Classification accuracy of every hashing method reported for each individual. C Annotation of major cell types performed using gene expression data and clustering. Classification accuracy of every hashing method reported for each cell type with SD values representing variation across 3 individuals. % mito—percentage of mitochondrial genes for each cell type