Supplementary Figure 1: Origin of peptide sensors, assay repeatability, and comparison of dynamic range to data variability across peptides.

(a) Computational curation established the connectivity between kinase enzymes and their substrate proteins, phosphorylatable target sites, and related peptide sequences ³⁰ (http://cancer.ucsf.edu/phosphoatlas; DOI: 10.1158/0008-5472.CAN-15-2325-T; US20120296880). (b) Projection of the coverage of biological peptides and generic positive control peptides per kinase. Generic CON⁺ peptides were systematically gathered from commonly available/advertised single-probe kinase assays/screens. (c) Schematic of the procedure. Details of the 228-peptide library we generated and used are available in Supplementary Tables 1–4. (d) Run-to-run comparison of kinases’ peptide-phosphorylation profiles demonstrating repeatability. (e-f) Example of JAK2’s peptide-phosphorylation signature. Four independent runs are analyzed. ‘Raw’ luminescence readouts are compared in (e) and Supplementary Table 5. Results transformed as internally normalized data by centering ATP-consumption profiles against the mean activity across all-228 peptides within each experiment, are shown in (f) and compared in Supplementary Table 6. The stacked bar graph in (f) shows that ATP-consumption profiles (y-axis) follow similar trends for each peptide probe (x-axis) across all runs: most bars follow the same activity trend of either “all medium” (around ‘0’, i.e. close to each experimental mean activity), or “all-high”, or “all-low”, for any individual peptide. This demonstrates run-to-run reproducibility, data consistency, and supports good correlation (>0.9) between experiments (see (d) and Supplementary Table 6). (g) Cumulative bar plots for other recombinant kinases we tested. Profiles support assay repeatability (within graphs) and kinases’ activity signature specificity/discernibility (between graphs; visually displayed in Fig. 1a). (h) Plot merging all kinase profiles from all experiments. (i-j) Dilution and time course validation. Commonly used generic CON⁺ peptides were interrogated to assess HT-KAM’s output for a set of kinases. Profiles match quality standards in the field. In (j): n≥3 biologically independent sample experiments per kinase; two-sided student t-test; box plot length: 25% and 75% of data, center line: median, whiskers 25% - (or 75% +) 1.5 x IQR). (k) Heatmaps of the top-20 peptides associated with highest measurable activities per kinase. For ~75% of kinases, biological peptides perform as better sensors of kinases’ activity than currently available/advertised CON⁺ peptides. (l) Comparisons of FDR-corrected t-test (top) and Z-factor profiles (bottom) across top activity-reporting peptides per peptide category. Results in (k-l) show that for many kinases, their best generic CON⁺ peptides have a good but not optimal ability to report on their enzymatic activity; performance is improved when using/including biological peptides; peptide sequences derived from biological substrate proteins are well suited to measure the phosphorylation activity of kinases.