Landscape of signaling responses across breast cancer cell lines. (A-C) Node-edge graphs of (A) MCF7 (A), BT-20 (B), and MDA-MB-453 (C) depicting induction of pERK (blue arrows) and pAKT (red arrows) by different ligands. The arrows’ thickness denotes the maximum induction across all time points and doses, normalized for each target across all cell lines and ligands. Gray arrows denote non-significant responses (less than two standard deviations above the control). Outer and inner circles represent basal measurements where the diameter is proportional to the logarithm of the expression level and shading to the logarithm of the phosphorylation level; data were normalized across all cell lines. pFGFR1-4 were not measured. Gray outer circles and labels denote expression levels below the detection threshold and their diameter denotes the minimal circle size. Lines show which ligands bind which receptors (Additional file 2: Table S2). (D-F) Fold-change in pERK (D-E) or pAKT (F) levels following exposure of cells to EGF (D, F) or HRG (E). Trajectories for six specific cell lines are highlighted in color; all other trajectories are represented in gray. (G-I) Average node-edge graphs for each breast cancer subtypes: TNBC (G), HER2amp(H), and HR + (I). Edges and nodes have the same meaning as in (A) except that circle size, intensity of shading, and edge width are proportional to the interquartile mean across lines of the subtype. (J) Fraction of cells of each subtype showing significant pERK or pAKT responses following exposure to ErbB, FGF, IGF/INS, or HGF ligands at 1 ng/ml. Wilcoxon’s rank-sum P-values are shown. (K) Distribution of the cell line mean fold-change following stimulation at 100 ng/ml of the same ligands as in (J). (L) Distribution of cell line median pathway bias of the same ligands as in (J).