Fig. 2: A high-resolution atlas detailing peripheral immune cell alterations by MP after surgery.

a Immune cells were clustered based on the expression of all phenotypic markers using an unsupervised bootstrapped clustering algorithm. The clusters were projected into two dimensions and major immune cell compartments were identified based on phenotypic marker expression (contoured in orange/green for innate/adaptive immune compartments, respectively). b Univariate p-values (two-sided Wilcoxon Rank Sum Test) were computed for each cluster at each time point to quantify the difference in functional marker expression or cell frequency between samples in the control (n = 30 patients) and MP (n = 28 patients) groups. At each time point, clusters were colored by the best univariate p-value observed for cell frequency and functional marker expression. c A Random Forest model was trained to classify patients in the control or MP group at each timepoint based on cluster-derived cell frequency and intracellular signaling responses. The boxplot depicts the probability predicted by the Random Forest model that samples from patients in the control (gray) or MP (blue) group were allocated to the MP group. The model revealed that samples from placebo- or MP-treated patients were distinguishable at 1 h (AUC = 0.91, p = 1.03E−7), 6 h (AUC = 0.92, p = 3.16E−8), 24 h (AUC = 0.85, p = 3.81E−6), and 48 h (AUC = 0.76, p = 3.2E−3) after surgery (two-sided Wilcoxon rank-sum test, p-values calculated for each unique model). All boxplots show median values, interquartile range, whiskers of 1.5 times interquartile range.