From diagnosis to therapy: algorithmic methodologies for precision medicine

In recent years, it has become established the idea of a novel medicine where a patient is the center around which multidisciplinary teams (made up of physicians, statisticians and bioinformaticians) sew targeted treatments. Precision medicine involves the use of detailed patient-specific molecular information for diagnosing, categorizing and guiding treatment of a disease, with the main purpose of improving the clinical outcome compared to a more classical approach. In precision medicine it is supposed that the cause of a disease is at least partially attributable to specific genetic or epigenetic characteristics of a patient. Therefore, identifying these specificities helps building the best treatment for each individual. Next-generation sequencing techniques are massively employed, giving the ability to quickly and at relatively low cost analyze whole genomes, epigenomes and transcriptomes. This ability is clinically important since the prediction of treatment effectiveness is usually affected by many factors. A fundamental function in this new medicine is played by bioinformatics. It has a crucial role in every aspect of precision medicine, such as the accurate classification of patients, the prediction of new therapies based on current knowledge, the identification of possible outcomes of a disease or therapy, and the enrichment of current knowledge on pathogenic processes or on pharmaceuticals. The aim of this thesis is the development of an integrated framework, based on synergistically operating tools, models and algorithms, which help to fill some of the major gaps in each step of the production of highly customized therapies, overcoming, if possible, the limitations of currently employed techniques, defining a new standard for precision medicine informatics.