AIM in Genomics: Ontological and connectivity structure of disease-gene modules in the human interactome

Image credit: [Lorenzo Madeddu]


The reductionist approach has dominated scientific research for several centuries and has been widely applied in the biomedical field. However, as we move into the realm of complex diseases, this approach fails to provide the insight needed to explain disease pathogenesis. Network medicine is a new paradigm that applies network science, artificial intelligence (in particular, machine learning and graph mining), and systems biology approaches to study a disease as a consequence of physical interactions within a cell. Pieces of evidence in this field show that if a gene or molecule is involved in a disease, its direct interactors might also be suspected to play some role in the same pathological process. This evidence has lead to formulating the so-called “disease module hypothesis”: genes involved in the same disease show a high propensity to interact with each other. However, the number and structure of disease modules are largely unexplored. The purpose of this study is to systematically analyze the relationship between structural proximity of disease modules and categorical similarity of diseases, by aligning human-curated disease taxonomies with disease taxonomies automatically induced from proximity relations of disease modules within the human-gene interaction network (interactome). We propose a large-scale analysis of a vast collection of diseases leveraging a novel network and taxonomy perspective. Our aim is to support clinical studies by obtaining relevant insights to improve our understanding of disease mechanisms at the molecular level.

By Springer Nature
Lorenzo Madeddu
Lorenzo Madeddu
Senior data scientist, PhD

Lorenzo Madeddu è senior data scientist (R&D) in the Knowledge Graph Insights team at AstraZeneca.