Predicting kinase activity and inhibitor sensitivity in MM by unbiased phosphoproteomics. (A) Schematic of the pipeline for kinase activity prediction from phosphoproteomic data. All phosphoproteomics were performed in biological triplicate and combined by averaging the log₂-transformed intensities of phosphosites associated with each kinase to generate activity scores. (B) Association of predicted FGFR3 activity from KSEA with known genetic aberrations. (C) Heatmap of the KSEA-predicted activities of 14 kinases that exhibited differential activity signatures across myeloma cell lines. The significance of the score from the median activity across cell lines was calculated by z-statistics (see “Methods”). *P ≤ .05, **P ≤ .01, ***P ≤ .001. (D) Viability curves showing the drug response of 8 myeloma cell lines to 8 kinase inhibitors (n = 4, mean ± standard deviation), with only INK128 and alisertib exhibiting strongly differential effects. (E) Correlation between inhibitor sensitivity and KSEA-predicted kinase activity for mTOR and aurora kinase A across myeloma cell lines shows modest predictive power. P values were calculated based on the null hypothesis that no relationship exists between the activity of a kinase and its sensitivity to an inhibitor. LC50, 50% lethal concentration; IMAC, immobilized metal affinity chromatography; LC-MS/MS, liquid chromatography-tandem mass spectrometry.