ROLE OF SOMATIC COPY NUMBER VARIATIONS IN CANCER

Genetic variation is the main reason of the phenotypic differences among individuals, as well as of many human genetic diseases. Recent advances in the methods to study the human genetic variation allow better identification of its different forms, in particular of copy number variations (CNVs). The causative role of germline CNVs in Mendelian diseases and in cancer predisposition is well established. Moreover, the driver role of cancer somatic CNVs is recently emerging, and large-scale quantitative analyses elucidating their functional role in cancer genomes are needed. To achieve this, we have analysed the genomic landscape of somatic CNVs in cancer genomes in comparison to germline CNVs in the genomes of healthy individuals. We observed that somatic CNVs substantially affect the genic portion of the genome, preferentially targeting cancer genes. Moreover, this is independent of genomic features, such as DNA repeating elements and recombination rate. In particular, we confirmed that oncogenes are preferentially amplified and tumour suppressors are preferentially deleted. To investigate their functional impact, we measured the gene expression changes upon copy number variation. We observed that amplification of a gene leads to its higher expression whereas deletion results in decreased gene expression, which suggests that amplifications activate dominant genes and deletions inactivate recessive genes. The two classes of cancer genes are vastly modified consistent with their functional roles as oncogenes and tumour suppressors, with the few exceptions of frequently amplified recessive genes underlying complex epigenetic regulation. The mutational spectrum of the human genes in cancer, together with their systems-level properties, can be exploited to identify novel targets for anti-cancer therapy, in which synthetic lethality emerges as a promising approach. Based on the working hypothesis that paralogous genes may engage in synthetic lethal interactions due to the functional redundancy between them, we combined several gene properties to predict synthetic lethality between paralogous gene pairs. Out of 37 candidate gene pairs, we experimentally validated the synthetic lethal interaction between two components of the cohesin complex, STAG1 and STAG2. Finally, we present the latest release of Network of Cancer Genes (NCG 5.0), a manually curated database of cancer genes and their systems-level properties. NCG 5.0 collects a list of 1,571 cancer genes mutated in 13,315 cancer samples and 24 primary sites from 175 published papers. NCG has been increasingly appreciated as a central resource for cancer genomics research, facilitating candidate prioritization for hypothesis testing and experimental planning in a wide range of studies.