Extended Data Fig. 7: In-depth analysis of the gene expression profile of MS1.

(a) Top 30 differentially expressed genes (among highly variable genes) when comparing MS1 against other CD14 + monocyte states (MS4 and MS2). (b) Dotplot showing enrichment of pathways (KEGG database) for upregulated genes in MS1 vs. MS2 (FDR < 0.1, edgeR exact test). n = 15,021 and 11,439 cells for MS1 and MS2, respectively. Sizes of circles are proportional to the number of gene hits in a set, whereas color represents the enrichment score of each gene set. (c) Heatmap showing the average expression of genes that are differentially expressed (FDR < 0.1, two-sided Wilcoxon rank-sum test) between all MS1 cells from each patient in the ICU-SEP cohort and all MS1 cells from each patient in the ICU-NoSEP cohort (n = 2,153 and 1,442 cells from 8 and 7 ICU-SEP and ICU-NoSEP patients, respectively). To specifically identify genes that discriminate the two patient populations, genes are filtered for expression in-group fraction > 0.4 and out-group fraction < 0.6. (d) k-selection plot to determine the number of components for non-negative matrix factorization (NMF). Dotted line indicates selected number of components for further analysis. (e) tSNE plot of MS1 cells (n = 15,021 cells) colored by patient cohort of origin. (f) Scaled TPM usage values of each gene module derived from NMF analysis. The top 20 genes in each module are shown. (g) Scatterplots showing correlation between mean gene module usage in MS1 cells and sequential organ failure assessment (SOFA) scores for Int-URO and URO patients (top row), and Bac-SEP and ICU-SEP patients (bottom row). Sample size (n) for each cohort is indicated in Fig. 1b. Significance of the correlation (Pearson r) was calculated with a two-sided permutation test. Line and shadow indicate linear regression fit and 95% confidence interval, respectively.