Fig. 2. Longitudinal analysis of drug response profiles.

a Heatmap showing the pairwise-Pearson correlation of cell lines using the z-normalized AUCs of the filtered drug profile. Top bar identifies the timepoint (Time) and Patient ID (PiD) using the color code denoted in the figure. b Volcano plot of the mean difference in the AUC (Time_x-T₀) by the log significance of the interaction determined from the linear model. The top 5 drugs are highlighted, in addition to components of the NACT regimen, which show an acquired resistance. Dotted lines show cut offs for ${-\log }_{10}\left(p-value\right)\ge 1.3\&\mathrm{abs}\left(mean\left(\mathrm{\Delta }AUC\right)\right)\ge 0.05$) used to prioritize the top compounds from the screen. c Heat map showing the shifts in AUC values for each matched pair. Y-axis shows the Patient ID number grouped by Mid-to-Pre and Post-to-Pre comparisons. X-axis is grouped according to pre-established mechanistic class and annotated by target.