Figure 1.

Discovering new drug biomarkers directly in clinical cancer genomics data sets (e.g., TCGA) using IDWAS. Data flow diagram describing the proposed novel methodology (IDWAS) for discovering novel pharmacogenomic variants. Boxes represent data, and circles represent processes. Predictive models of drug response are fit in preclinical data (in this case GDSC cancer cell lines) and applied to tumor gene expression in a clinical cancer data set (in this case TCGA). These imputed drug response data are then associated with measured somatic variants in the clinical data set in order to discover novel biomarkers of drug response.