Gene multiplex network integration

AGUSTIN SOMACAL; ARIEL CHERNOMORETZ

Taller; Machine Learning Summer School; 2018

Data integration is becoming a fundamental challenge in many areas as huge sources of information are available. This is particularly evident in human genetics, specially when trying to answer the question of which are the relationships between genes, proteins and hereditary diseases. Several state of the art techniques make use of prioritization algorithms over networks extracted from a single data type, generally a protein-protein interaction network (PPI) . However, it is possible to build multiplex networks where nodes represent proteins but relationships of different nature between them are captured considering different layers of edges. In this work we propose a novel approach to build a weighted version of the PPI network that captures connectivity information of an initial multiplex network, better suited for this kind of disease gene prioritization problems.