Fig. 1

Sample multiplexing using DNA-barcoded antibodies. a Schematic overview of sample multiplexing by Cell Hashing. Cells from different samples are incubated with DNA-barcoded antibodies recognizing ubiquitous cell surface proteins. Distinct barcodes (referred to as hashtag-oligos, HTO) on the antibodies allow pooling of multiple samples into one scRNA-seq experiment. After sequencing, cells can be assigned to their sample of origin based on HTO levels (“Methods” section). b Representative scatter plot showing raw counts for HTO A and HTO B across all cell barcodes. Both axes are clipped at 99.9% quantiles to exclude visual outliers. c Heatmap of scaled (z-scores) normalized HTO values based on our classifications. Multiplets express more than one HTO. Negative populations contain HEK293T and mouse NIH-3T3 cells that were spiked into the experiments as negative controls. d tSNE embedding of the HTO dataset. Cells are colored and labeled based on our classifications. Eight singlet clusters and all 28 cross-sample doublet clusters are clearly present. e Distribution of RNA UMIs per cell barcode in cells that were characterized as singlets (red), multiplets (violet) or negatives (grey). f Transcriptome-based clustering of single-cell expression profiles reveals distinct immune cell populations interspersed across donors. B, B cells; T, T cells; NK, natural killer cells; mono, monocytes; DC, dendritic cells. Cells are colored based on their HTO classification (donor ID), as in d