Extended Data Fig. 2: Protein expression levels and IFN-I network connectivity across PBMC subsets.

a. Single-cell heatmap showing the expression of proteins used for clustering in each of the clusters defined by fast-Phenograph. Proteins and clusters were ordered by hierarchical clustering, white spaces divide distinct clusters and expression is scaled by row for each protein and expressed as a percent of maximum expression. b. Kaplan–Meier curves showing overall survival and progression-free survival comparing patients stratified based on their IS in the indicated PBMC population. Sample sizes for each group are shown on the plots. c. Pairwise correlations between median IS across major immune populations. The thickness and color of the lines connecting each population signifies the strength of the correlation. A significant value greater than 0 indicates that higher IS in one subset is accompanied by higher IS in the subset it is compared to, with 1 indicating a perfect correlation. Correlations are overlaid on UMAP coordinates from Fig. 1b. Only correlations with P < 0.05 are shown. P value determined by two-sided Student’s t-test. d. Network map showing the difference in correlation coefficient between patients with no clinical benefit of treatment (NCB, n = 18) and clinical benefit (CB, n = 13) in pre-therapy samples. The thickness and darkness of the line indicates the magnitude of the difference and the color indicates the direction (Red = higher in NCB; Blue= higher in CB). Only differences greater than 0.2 in either direction are shown. e. Same as (c), except mapping the correlations in IS for patients after therapy initiation. f. Same as in (g), except mapping the change in correlation between post-treatment (6 weeks) and pre-treatment for NCB (left) or CB (right) patients.