Fig. 1: Overview perturbation-based benchmark.

a Workflow for kinase activity inference. Kinase activities can be inferred from phosphoproteomics data when combining it with a kinase-substrate library linking each kinase to its downstream targets and an inference method. b Overview of perturbation experiments. In total, 230 experiments with known perturbation targets where either an increase or decrease of the kinase activity is expected are included in the benchmark. c Overview of metrics for measuring the performance. P_Hit(k) calculates the probability of predicting the perturbed kinase among the top k kinases based on the inferred activities. The scaled rank calculates the rank of the perturbed kinase divided by the total number of kinases in each experiment. The area under the receiver operator characteristic curve (AUROC) is calculated by ranking kinases across all experiments based on inferred activities, with perturbed kinases as true positives and all other kinases as true negatives. d–f Evaluation of kinase activity inference methods using the d probability of predicting the perturbed kinase among the top k (5, 10, 20) kinases based on the inferred activities, e the scaled rank of the perturbed kinase activity (n = 37) for each experiment and f the AUROC of kinases ranked across experiments by their activities. The AUROC calculation was repeated a thousand times, with randomly selecting a subset of the negative classes with the same size as the positive class (n = 1000). For the boxplots, the central line depicts the median, the box hinges represent the 25th to 75th percentiles, and the whiskers extend up to 1.5 times the interquartile range above and below the box hinges. Outliers are depicted as individual hollow points beyond the whiskers. Source data are provided as a Source Data file.