Fig. 1

PaSSS analysis of TNBC patient-derived samples. A Steps of the analysis: Proteins whose phosphorylation levels deviate from the reference (steady) state in the same or opposite directions (left panel) are identified via co-variance matrix and surprisal analysis (exemplified by pEGFR, ER and pSrc, right panel). Here pEGFR and ER deviate in the same direction from the steady state, whereas pEGFR and pSrc are orthogonal. B Protein weights (Gi) are quantified (Supplementary Methods). Proteins located on the tails of Gi plots are co-expressed and grouped into unbalanced processes using STRING protein–protein connections and Gi. For each sample, PaSSS is assigned to represent only sample-specific processes, e.g., with significant amplitude \({\lambda }_{\alpha }\left(k\right)\) (all steps of the analysis are detailed in Supplementary Methods). C-D Processes found in BR45, BR98 and 613 TNBC subgroup: see process 1, full in (C) and zoomed in in (D). Zoomed in images of processes 1 and 2 representing this subgroup are shown in (D). Proteins with multiple phosphosites include MUC1 (YYYYY: pS1207-pY1212, pY1212, pY1203, pY1209 and pY1209-pY1212), EGFR (YY: pY1172 and pY1197) in process 1, and EGFR (YY: pY1172 and pY1197) in process 2. E 2 processes were sufficient to reproduce this subset and thus provide the full characterization of the data (Supplementary Methods). F Mapping of BR45, BR98 and 613 tissues in 2D space using \({\lambda }_{\alpha }\left(k\right)\) values. The sign of \({\lambda }_{\alpha }\left(k\right)\) determines the location on the map in terms of sample separation in nD space (Supplementary Methods). Sample location in (F) is schematically converted into a “barcode” (G). Red-labeled proteins in (D) are upregulated in the red-labeled processes in “PaSSS” barcodes (G), whereas blue-labeled proteins are upregulated in blue-labeled processes (Supplementary Methods)