The amount of genomic data produced by DNA-sequencing is growing at an unprecedented rate due to the ever greater throughput provided by the new generation of genome sequencing (NGS) platforms. To understand and interpret biological data a huge number of metadata and annotations are required. Genomic metadata include very heterogeneous biological and clinical attributes gathered at different levels of details such as diseases, genes, proteins, interactions, pathways but also phenotypic characterization of patients and clinical evidence or diagnosis. All these information are required to draw a comprehensive picture of many underlying phenomenons, thus contributing to scientifically understand the observed data. The complexity of genomic data arises due to the number of involved entities (from millions to billions) and the complex relationships between them; biological information is typically highly connected, not uniform, semi-structured and unpredictable. Relationships and connections may be stored in a relational database and data can be extracted adopting traversal-type queries implying joins. Nevertheless joins with large tables easily become too cumbersome and computationally expensive to design, execute and maintain. This critical aspect makes relational databases non-suitable for this kind of data operations. Studies already suggested that graph databases are among the best choices to explore linked data, due to the design of the core engine which optimizes performance in exploring connections [1]. They also provide a flexible solution for the integration and exploration of multiple levels of biological information [2] and some specific sets of biological data are already structured as graphs [3].