Systematic studies of cancer genomes are providing unprecedented insights into the molecular nature of human cancer. Using this information to guide the development and application of therapies in the clinic is challenging. We have performed a large-scale integrative study of pharmacogenomic data encompassing the characterisation of 1,001 cancer cell lines and 11,289 primary tumours from 29 different tissues. Results from this analysis show that cancer-driving alterations identified in primary tumors (integrating mutations, copy-number alterations, methylation and gene expression) are informative of the response to 265 compounds profiled in human cancer cell-lines. Additionally, cell-lines recapitulate partially but faithfully the landscape of oncogenic aberrations identified in tumors, and many of these aberrations interact statistically with drug sensitivity or resistance. Finally, logic-based modeling uncovers combinations of aberrations that specifically sensitize to drugs, and machine-learning techniques can be used to explore the ability of different data omics (and their combinations) in predicting drug response. In my talk, I will illustrate how these analyses were performed, showing representative results and proposing associated datasets and web-portals as possible resources for the identification of novel therapeutic options for selected sub-populations of cancer patients.