Harnessing synthetic lethality — which occurs when the inhibition of two non-essential genes is lethal — is attracting significant attention as a strategy to selectively treat cancer. Here, the authors use the data mining synthetic lethality identification pipeline (DAISY) to analyse the accumulating cancer genomic data. They construct a genome-wide network of synthetic lethality interactions in cancer, which they demonstrate to successfully predict gene essentiality, drug efficacy and clinical prognosis.