Fig. 1: Identification of consensus endotypes in public data.

a, Peripheral blood gene expression data from 19 cohorts inclusive of 548 samples from bacterially infected patients and 912 samples from virally infected patients co-normalized. We calculated the five sepsis signatures and scaled values for unsupervised clustering. b, Unsupervised hierarchical clustering performed by scaled gene expression score (x axis) across all samples (y axis) identifying four consensus endotypes. c, The four identified consensus endotypes separated well in PCA. d, Network analysis performed on scaled scores using Spearman’s correlation >0.33 to identify edges. Clusters were identified using a greedy forward algorithm, which identified four clusters mirroring those identified by unsupervised hierarchical clustering. The thickness of the line represents correlation between the nodes it connects. Coag, coagulopathic; Inflamm., inflammopathic; Mod, module.