Phd in DATA SCIENCE

Thesis title: Network-Based Algorithms for Biological Data Analysis

Diseases rarely result from anomalies within single genes. Rather, they reflect perturbations of the complex intracellular and intercellular network that links tissue and organ systems. Network medicine offers a general approach to systematically address the complexity of a particular disease, leading to the identification of new pathways, the discovery of new drug targets for disease treatment, and the molecular relationships between different phenotypes. Advances in network medicine can help identify new disease genes, advance our understanding of disease-associated mutations, discover new bio-markers for complex diseases. In this line of research, we designed new algorithmic solutions in the field of network medicine, investigating the problems of disease-gene prioritization and link prediction in the Protein-Protein Interaction network (PPI). As a first contribution, we developed the Biological Random Walks (BRW) framework for disease-gene prioritization. The proposed framework leverages the integration of multiple biological sources within a propagation-based approach, integrating biological information across connected nodes in a given network. As a second contribution, we investigated the problem of identifying causal genes involved in known disease variants. In this case, we designed a new framework that leverages the topology of a suitable co-regulation network to prioritize one causal gene per disease variant. Finally, investigated the problem of predicting candidate protein pairs involved in pairwise interaction in the PPI, a key task to achieve increasingly accurate descriptions of the interactome. To this purpose, we leveraged the similarity of protein pairs using both network topology and the protein's primary structure, designing a framework that combines them to predict new candidate interactions.