Abstract Human cancers are highly heterogeneous diseases that rely on the accumulation of genetic and epigenetic alterations for initiation and maintenance. The increasing knowledge emerging from large sequencing efforts, such as The Cancer Genome Atlas (TCGA) or the International Cancer Genome Consortium (ICGC) can be used to identify new targets for treatment by applying the concept of synthetic lethality (SL), with new synthetic lethal pairs identified through mutual exclusivity analysis of patient-derived tumor data. However several confounding factors, such as sample purity or sequencing platform, prevent the systematic explorations of the full spectrum of synthetic lethal (SL) pairs. Here, we present a two-step approach to handle large SL studies: i) SPICE (Synthetic Lethal Phenotype Identification through Cancer Evolution analysis) pipeline to process in a uniform environment thousands of samples; (ii) FaMe (Fast Mutual Exclusivity algorithm) to explore millions of aberration pairs and then nominate mutual exclusive pairs, which will be validated in vitro. The SPICE pipeline handles DNA sequencing data (whole genome, whole exome, and targeted sequencing) and generates purity- and ploidy-adjusted somatic copy number profiles. Then, it assesses allele specific copy number, a strong confounding factor in mutual exclusivity analysis. For instance, a copy number neutral LOH gene is interpreted as “wild type” in classic log2 ratio-based copy number analysis, potentially impairing SL analysis. Moreover, SPICE pipeline distinguishes between clonal and subclonal copy number events and processes single nucleotide variants (SNVs) and structural rearrangements. Afterwards, FaMe models each type of aberration in a dataset as a binary matrix reporting whether a gene in a sample is aberrant or not. The approach is quite general and searches for any combination of aberration types (SNVs, deletions and amplifications). Fast matrix multiplication, logarithm-based implementation of Fisher's exact test and massive parallelization allow testing for mutual exclusivity of hundreds of millions of aberration pairs in few minutes. Filters based on expression data, interaction databases, and drug responses resources are integrated to shortlist nominated pairs for in vitro validation.We processed 8,397 samples belonging to 29 tumor types from TCGA. Preliminary results identify known SL pairs as for example the mutual exclusivity of SNVs and combination of homozygous deletion and SNVs of the EGFR and K-RAS genes in lung adenocarcinoma (LUAD) (p= 0.005 and p= 0.0029, respectively). Details on the general features of these analyses and on the newly identified mutually exclusive pairs will be presented. Citation Format: Francesca Lorenzin, Davide Prandi, Paola Gasperini, Tarcisio Fedrizzi, Francesca Demichelis. Synthetic lethal pairs for cancer treatment by agnostic exploration of mutually exclusive genomics [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr LB-022.