Patient-centric data integration framework for highly dimensional data
Host: BARCELONA SUPERCOMPUTING CENTER (Spain)
PhD awarding institution: UNIVERSITAT DE BARCELONA
Lead Supervisor: N Pržulj
Objectives: PM proposes to individualize the practice of medicine based on patients’ genetic backgrounds, their biomarker characteristics and other omics datasets including exposure. ESR9 will build upon our previous work on network science, data integration and PM to propose a patient-centric data integration framework that enables all of the following: (1) improved patient stratification (allowing for predicting disease outcomes with more confidence), (2) uncovering molecular bases of diseases (molecular mechanisms, disease genes, biomarkers), and (3) personalized treatment predictions (drug repurposing). In practice, the mutational data will be mapped onto molecular networks and graphlet-based approaches will be utilized for mining for medically relevant signals. All data will be integrated using non-negative matrix factorization based approaches (Zitnik et al 2013; Gligorijevic et al. 2016) into a unified framework from which additional knowledge will be mined. Unlike existing approaches that only represent and integrate biological data as networks, we will thus also consider alternative data representation, such as hyper-networks and simplicial complexes that can capture the multi-scale organization of the data.