Supplementary Figure 2: Including a multiplicity of peptide sensors considerably improves the performance of any enzyme activity assay designed to best differentiate and identify kinases.

(a-b) Profiles of AUC values calculated for increasing numbers of randomized combinations of peptide sensors (1 to 50 in (a); 1 to 100 in (b)). (c-e) Kinase phosphorylation activity measured with single generic CON⁺ peptides. In (c), activity levels compare what is ‘expected’ from advertised single-peptide assays (top panel), to what is experimentally measured (bottom panel; excerpt from kinases’ 228-peptide phospho-activity profiles). In principle, one may expect that the top red/blue kinase-peptide intersections would match the bottom ‘redder’/‘bluer’ patterns. In (d), percent concordances between expected and experimental activities are shown (average 52%). Kinases/peptides ‘cross-reactivity’ observed in (c-d) indicates that no single generic CON⁺ peptide would allow to accurately identify and differentiate a particular kinase from other kinases. In (e), we calculated the single-peptide AUC values of the 14 generic CON⁺ peptides to evaluate how specific/sensitive they are at identifying their respective kinases. The bottom panel displays AUCs for 14 individual biological peptides with highest measurable AUC for each kinase. Most single generic CON⁺ peptides do not provide highest possible specificity/sensitivity for kinases they are expected to report on. Single biological peptides vastly outperform single generic CON⁺ peptides. Although this does not undermine the utility of single generic CON⁺ peptide-based assays for pharmaceutical screens, this underlines why a multi-peptide approach is a valuable alternative to single-peptide measurements when investigators want to specifically and differentially identify a kinase from others. (f) ROC curves and AUC values of phospho-catalytic activity profiles established with distinct peptide sets: kinases’ biological peptides, kinases’ generic CON⁺ peptides, random peptides. (g-i) Combinatorial peptide sets that best differentiate between enzymatic signatures of kinases. Differential signatures (g-h) and their AUCs (i) were computed. The differential signatures of EGFR and SRC included 42 and 43 peptides (AUC’s: 0.840, 0.893) and passed p<0.05 for FDR-corrected t-test or Wilcoxon rank sum test (but not both concurrently). (j) AUCs measured with all-biological, all-positive, all-random, or all-differential peptide sets across all kinases/kinase families. The specificity/sensitivity provided by biological peptide subsets performed almost to the same degree as differential peptide pools, and significantly outperformed generic CON⁺ peptides. In the perspective of expanding the coverage of the platform to monitor more/other kinases, the computational analyses and modeled data in (a-b,f-j) and Fig. 1c–f demonstrate that including additional peptide probes (especially biological peptides) would only increase the capability of the HT-KAM-screen to accurately predict the identity of more/different kinases using their phospho-catalytic signatures.