Design and evaluation of bioinformatic tools to detect rare CNVs

In recent years we have witnessed to an increasing interest in the development of computational tools to study genetic mutations. Many bioinformatic tools have been developed in order to statistically classify and detect such mutations from large amounts of biological data. In this thesis we develop and test a novel pipeline to solve the problem of rare germline CNV detection. We adopt an approach based on a train/validation paradigm in order to achieve an optimal setting of the parameters of the pipeline. The pipeline has been tested on real data coming from a genetic company based in Padova in order to assess its performance in a real world setting.