Fig. 2: Defective IFN-γ response and reduced transition potential to ncMono in monocytes of severe COVID-19.

a, UMAP embedding of 116,944 monocytes and DC. Seven cell types were defined by RNA expression of marker genes (Extended Data Fig. 2a). b, Graph representation of Nhoods identified by Milo. Nodes are Nhoods, colored by their log₂ FC between COVID-19 (n = 72) and healthy controls (n = 75) adjusted by age and sex. Nondifferential abundance Nhoods (FDR ≥ 0.1) are colored white. c, Beeswarm plot showing the distribution of adjusted log₂ FC in abundance between COVID-19 and healthy controls in Nhoods according to seven cell types. Colors are represented in the same way as in b. d, The top ten enriched biological processes by GO analysis of upregulated DEGs of moderate and severe disease compared to healthy group in five cell types. Dot color indicates the statistical significance of the enrichment (adjusted P values via the Benjamini–Hochberg method), and dot size represents gene ratio annotated to each term. e, The differential gene expression analysis between moderate (n = 8) and severe (n = 64) COVID-19 in ncMono. DEGs (FDR < 0.05 and FC > 2) are colored in light blue and labeled by gene symbols if log₂ FC > 1.5. f, Velocities derived from the dynamical model for monocytes and DC cluster from COVID-19 and healthy group are projected into a UMAP-based embedding. Colors indicate the same clusters as in a. g, Average unspliced ratio of each sample stratified by three monocytes clusters, colored by COVID-19 (n = 73) and healthy (n = 75) groups. Condition-specific regression lines are shown. P value for the interaction effect between three monocyte clusters and two clinical conditions is uncorrected and reflects two-sided test.