Extended Data Fig. 6: SARS-CoV-2-RNA⁺ cells distinguished by sc/snRNA-Seq.

a, Detection of SARS-CoV-2 UMIs from sc/snRNA-Seq data. SARS-CoV-2 UMIs from all cell barcodes (top) and after ambient correction (second from top). Number (second from bottom) and percentage (bottom) of SARS-CoV-2 RNA⁺ cells after ambient correction (m = 24 specimens). b, c, Effect of ambient RNA on SARS-CoV-2 RNA⁺ detection. Number of SARS-CoV-2 aligning UMI per cell barcode (CB) (y axis) in healthy lung (b, black), in vitro SARS-CoV-2 infected human bronchial epithelial cells (HBEC)⁵⁶ (b, blue) or lung samples from COVID-19 donors at autopsy either with CB with high-quality capture of human mRNA (b, red) or after removal of cells whose viral alignments were attributed to ambient contamination (c, Supplementary Methods). d, Variation in SARS-CoV-RNA⁺ cells across donors. Percentage of cells (y axis) assigned as SARS-CoV-2 RNA⁻ (white), SARS-CoV-2 RNA⁺ (red) or SARS-CoV-2 ambient (grey, Supplementary Methods) across the donors (x axis), sorted by proportion of SARS-CoV-RNA⁺ cells. e–i, Viral RNA detection does not correlate with cell quality metrics. e–h, Number of SARS-CoV-2 UMIs (before ambient viral correction) for each cell (y axis) versus either number of SARS-CoV-2 genes for that cell (e, x axis), number of human (GRCh38) genes per cell (f, x axis), number of human (GRCh38) UMI per cell (g, x axis) or percentage of human (GRCh38) mitochondrial UMIs per cell (h, x axis). i, Number of retained high-quality cells (x axis) and number of SARS-CoV-2 RNA⁺ cells (y axis) in each sample (dots) after correction for ambient viral reads. Pearson’s r = 0.07, two-sided P = 0.73. j–l, Agreement in viral RNA detection between qPCR and sn/scRNA-Seq. Number of SARS-CoV-2 copies measured by CDC N1 qPCR on bulk RNA extracted from matched tissue samples (x axis) and the number of SARS-CoV-2 aligning UMI (y axis) for each sample (dot) from all reads (j, P < 0.0001, two-sided), all reads from high-quality cell barcodes (k, P < 0.0001), and after viral ambient RNA correction (l, P = 0.0042). Spearman’s ρ reported, two-sided test. m, Genetic diversity of SARS-CoV-2. Maximum likelihood phylogenetic tree of 772 SARS-CoV-2 genomes from cases in Massachusetts between January and May 2020. Orange points, donors in this cohort. n, Specificity of SARS-CoV-2 infection. log₁₀(1+ reads) in each donor (columns) assigned to different viruses (rows) by metagenomic classification using Kraken2 from bulk RNA-Seq. Asterisks denote targeted capture. o–u, Relation between SARS-CoV-2 RNA and different cell types. Number of SARS-CoV-2 aligning UMIs in each (including all CB) and the proportion of epithelial (o), mast (p), macrophage VCAN^highFCN1^high (q), macrophages CD163^highMERTK^high (r), macrophages LDB2^highOSMR^highYAP1^high (s), venular endothelial (t) or capillary aerocytes (u) cells in these samples (x axes). Pearson’s r denoted in the upper left corner with significance after Bonferroni correction (P). v, Effect of viral load on bulk RNA profiles. Significance (−log₁₀(P), y axis) and magnitude (log₂(fold change), x axis) of differential expression of each gene (dots) between three donors with highest viral load and six donors with lowest or undetectable viral load profiled by bulk RNA-Seq. Red points, FDR < 0.05. w–y, Distribution of SARS-CoV-2 RNA⁺ cells across cell types and subsets. Number of SARS-CoV-2 RNA⁺ cells (y axis) from each donor (colour) across major categories (w, x axis), myeloid subsets (x, inflammatory monocytes: 40 cells, five donors; LDB2^highOSMR^highYAP1^high macrophages: 27 cells, five donors; x axis), or endothelial subsets (y, capillary endothelial cells: 16 cells, four donors; lymphatic endothelial cells: nine cells, three donors; x axis).

Source data