Fig. 2

Demonstration of biomarker pairs for response analysis using TMB (marker1) and the TGF-beta signature (marker2). Boxplots display the distribution of the biomarkers (left: TMB; right: TGF-beta) by response group (positive vs negative); Wilcoxon tests were employed to compare the response group for each biomarker (a). The scatterplot displays the inter-marker correlation with the color representing the response category, and red indicates positive (b). The ROC curve depicts the performance of the single marker and dual marker using a logistic regression model with or without marker interaction. The AUC and its confidence interval are presented on the plot as well (c). Samples were stratified into 4 groups/quadrants, and statistics for each group are shown (d–g). The area proportion chart shows the group size and independence test of two markers by Fisher’s exact test (d). The quadrant statistic matrix presents the group size, response rate and confidence interval (e). The doughnut plot illustrates both the relative group size (width of ring) and response rate (red arc fraction, f). The line chart shows the response rate in each quadrant/group and the potential statistical interaction between two markers, i.e., if two lines are crossed (g). All figures were generated by dualmarker (v0.1.0)