Skip to main content
Fig. 3 | Genome Biology

Fig. 3

From: Application of single-cell RNA sequencing in optimizing a combinatorial therapeutic strategy in metastatic renal cell carcinoma

Fig. 3

Identification of targetable signaling pathways by transcriptome profiling and drug screening. a Quantitative estimation of the activation status of targetable signaling pathways across single cells. Boxplots demonstrate overall reciprocal differences in expression signatures across normal kidney cortex, bulk cells of each population, and single cells. b Measured drug response profiles of pRCC and mRCC cells, matched to the targetable signaling pathways. Sensitivities of cells to various targeted drugs were determined based on the half-maximal inhibitory concentration (IC50), and transformed to Z-scores. Afatinib and dasatinib were selected as the most effective drugs against mRCC cells (denoted as *) whereas everolimus and pazopanib (denoted as †) showed no effects, which is consistent with clinical findings. ce Drug sensitivity was predicted by the ridge regression model using a training set of publicly available cancer cell line expression data with each of the measured IC50 data. Estimated values were transformed to Z-scores across samples. c Significant correlation of predicted drug sensitivity with measured sensitivity in b. d Comparison of the predicted drug sensitivity of afatinib and dasatinib between populations. e For the selected drugs afatinib and dasatinib, there was a significant correlation between predicted drug sensitivity (Z-scores) and activation status (GSVA scores) of the relevant targeted pathways. c, e Linear regression was applied to estimate Pearson’s correlation coefficient (r), with 95 % confidence as shown by thicker light gray curves. The statistical significance of the regression was determined by one-way ANOVA test. a, d Boxes show 25th to 75th percentile with 10th and 90th percentile whiskers. *P <0.05, **P <0.01, ***P <0.001, two-tailed Student’s t-test

Back to article page
\