Fig. 4: Combining kinase-substrate libraries to optimize kinase activity inference.

a–c Kinase activity inference performance (AUROC) using the a perturbation-based, b activating site-based and c protein-based benchmark approach when considering the target set combining known targets from the curated combination (PhosphoSitePlus, PTMsigDB and GPS gold) with targets from OmniPath, NetworKIN or iKiP-DB. The performance was once evaluated when considering all possible kinases for each library (right) and once evaluated after filtering for kinases that could be evaluated using kinases considered in the curated combination. 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 for the perturbation-based benchmark and 80% of the gold standard set of the tumor-based benchmark (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.