Abstract Motivation: Synthetic Lethal (SL) interactions can provide a basis for the development of personalized anticancer drugs with higher therapeutic efficacy and reduced toxicity. Emerging screening technologies have enabled the detection of SL interactions in human cell lines in recent years. However, such screens are confined to find the SL-partners of a small set of genes or drugs, and yet cannot encompass the large spectrum of cancer types and genetic interactions. As these screens are mostly conducted in cancer cell lines or in animal models, they may fail to capture the SL interactions that occur in human cancers in-vivo. As a result, the promise of harnessing SL interactions to develop selective anticancer drugs is yet to be fully explored and exploited. Methods: Here we present a novel computational framework for the genome-scale identification of SL interactions in cancer, which spans different cancer types in a clinical setting. Our approach stems from the basic assertion that cancer cells that have lost two SL-paired genes are strongly selected against. This observation enables us to identify SL interactions by analyzing the genetic profiles of thousands of clinical samples and detecting events of gene-co-deletions that occur significantly less than expected. Based on converging evidence from such an analysis of different genome-scale cancer datasets we have assembled a core set of SL interactions that are supported by at least three different datasets. These SLs are shared by the wide range of cancer types in a statistically enriched manner, and constitute the Cancer Synthetic Lethal Network (CSLN). Results: Testing the veracity of the CSLN, we employed it to infer gene essentiality in 26 different cancer cell lines, based on their genetic profiles. The obtained cell line-specific gene essentiality predictions significantly match pertaining experimental essentiality measures (Marcotte et al., Cancer Discovery, 2011). Remarkably, the significant predictive power of the CSLN surpassed that obtained when using a cohort of SL-interactions detected experimentally. Following, we comprehensively tabulated the therapeutic potential of targeting the CSLN genes in different cancer types, to provide an array of novel cancer-specific drug target and repurposing predictions. Extending the concept of synthetic lethality to target oncogenes that are difficult to target directly, we identified synthetic lethality arising from gene amplification, rather than deletion, and constructed the Cancer Amplification SL Network (CASLN). We employed it to predict the efficacy of 30 different drugs across 593 cancer cell lines, which have been recently tested experimentally (Garnett et al., Nature, 2012). Indeed, drugs were significantly more effective in the predicted sensitive cell lines than in those predicted insensitive. Significance: This study paves the way for obtaining and exploiting a system-level understanding of SL interactions in cancer in three important ways: 1. It provides a novel computational pipeline for detecting SL interactions in cancer. An approach that can also be employed to detect context-dependent SL interactions that are specific to certain cancer types (as we demonstrate by generating and testing a breast cancer SL network). 2. It presents two large-scale networks of genetic interactions in cancer, whose utility is manifested by their ability to predict gene essentiality and drug efficacy in a cell-line-specific manner. 3. The broad set of novel cancer-specific drug target predictions obtained is likely to provide a basis for numerous follow-up experimental studies. In summary, our approach opens up exciting opportunities for rational drug design and for the development of personalized therapeutic strategies for cancer. Citation Format: Livnat Jerby Arnon, Adam Weinstock, Tamar Geiger, Eytan Ruppin. Systematic reconstruction of the cancer synthetic lethal network and its application for the identification of selective cancer drug targets. [abstract]. In: Proceedings of the AACR Precision Medicine Series: Synthetic Lethal Approaches to Cancer Vulnerabilities; May 17-20, 2013; Bellevue, WA. Philadelphia (PA): AACR; Mol Cancer Ther 2013;12(5 Suppl):Abstract nr A32.