Müller-Dott Sophia, Jaehnig Eric J, Munchic Khoi Pham, Jiang Wen, Yaron-Barir Tomer M, Savage Sara R, Garrido-Rodriguez Martin, Johnson Jared L, Lussana Alessandro, Petsalaki Evangelia, Lei Jonathan T, Dugourd Aurelien, Krug Karsten, Cantley Lewis C, Mani D R, Zhang Bing, Saez-Rodriguez Julio
Heidelberg University, Faculty of Medicine, and Heidelberg University Hospital, Institute for Computational Biomedicine, Bioquant, Heidelberg, Germany.
Lester and Sue Smith Breast Center, Baylor College of Medicine, Houston, TX, USA.
Nat Commun. 2025 May 22;16(1):4771. doi: 10.1038/s41467-025-59779-y.
Kinases regulate cellular processes and are essential for understanding cellular function and disease. To investigate the regulatory state of a kinase, numerous methods have been developed to infer kinase activities from phosphoproteomics data using kinase-substrate libraries. However, few phosphorylation sites can be attributed to an upstream kinase in these libraries, limiting the scope of kinase activity inference. Moreover, inferred activities vary across methods, necessitating evaluation for accurate interpretation. Here, we present benchmarKIN, an R package enabling comprehensive evaluation of kinase activity inference methods. Alongside classical perturbation experiments, benchmarKIN introduces a tumor-based benchmarking approach utilizing multi-omics data to identify highly active or inactive kinases. We used benchmarKIN to evaluate kinase-substrate libraries, inference algorithms and the potential of adding predicted kinase-substrate interactions to overcome the coverage limitations. Our evaluation shows most computational methods perform similarly, but the choice of library impacts the inferred activities with a combination of manually curated libraries demonstrating superior performance in recapitulating kinase activities. Additionally, in the tumor-based evaluation, adding predicted targets from NetworKIN further boosts the performance. We then demonstrate how kinase activity inference aids characterize kinase inhibitor responses in cell lines. Overall, benchmarKIN helps researchers to select reliable methods for identifying deregulated kinases.
激酶调节细胞过程，对于理解细胞功能和疾病至关重要。为了研究激酶的调节状态，已经开发了许多方法，用于使用激酶 - 底物文库从磷酸化蛋白质组学数据推断激酶活性。然而，在这些文库中，很少有磷酸化位点可归因于上游激酶，这限制了激酶活性推断的范围。此外，不同方法推断出的活性有所不同，因此需要进行评估以获得准确的解释。在这里，我们展示了benchmarKIN，这是一个R软件包，能够对激酶活性推断方法进行全面评估。除了经典的扰动实验外，benchmarKIN还引入了一种基于肿瘤的基准测试方法，利用多组学数据来识别高活性或低活性激酶。我们使用benchmarKIN评估激酶 - 底物文库、推断算法以及添加预测的激酶 - 底物相互作用以克服覆盖限制的潜力。我们的评估表明，大多数计算方法表现相似，但文库的选择会影响推断出的活性，手动策划的文库组合在重现激酶活性方面表现出卓越的性能。此外，在基于肿瘤的评估中，添加来自NetworKIN的预测靶点进一步提高了性能。然后，我们展示了激酶活性推断如何有助于表征细胞系中激酶抑制剂的反应。总体而言，benchmarKIN帮助研究人员选择可靠的方法来识别失调的激酶。
