From genotype to phenotype: novel methodologies for the discovery of new pathological biomarkers in human

The DNA is the code that programs the biological processes of many organisms. It contains specific regions, e.g. genes, composed by sequences of nucleotides communicating between them or with other biomolecules, called regulatory elements, in order to accomplish many tasks. Pathological conditions arise from the erroneous functioning and regulation of these elements. Chapter 1 provides an overview of the topics covered in this thesis. In Chapter 2 are reported transcriptome analysis techniques related to microRNAs in Papillary Thyroid Carcinoma and to long non-coding RNAs in Diffuse Large B-cell Lymphoma. Chapter 3 explains the debated competing endogenous RNA mechanism of post-transcriptional regulation and presents CERNIA, a machine learning framework for predicting RNA crosstalk mediated by miRNAs. There are many approaches for retrieving mutual protein-protein interactions from gene expression. Protein expression could represents a useful although expensive resource. Chapter 4 proposes a boosting methodology that relies on several inference methods in order to improve the prediction accuracy of protein interactions along multiple human cancers and to shed light on their differences and commonalities. Finally, Chapter 5 introduces the field of radiogenomics, a promising approach studying the relationship between the imaging and genome-related characteristics of a disease, and provides the preliminary results of a study on patients with dementia and Alzheimer's disease that underwent 18F-Florbetaben PET.