Large-scale knowledge graph representations of disease processes

Abstract

Today, a wide range of technologies and data types are available when studying disease-relevant processes. Therefore, a major challenge is integrating data from different technologies covering different levels of functional cellular organization. This motivates approaches that start with a bird's-eye perspective, initially considering as many molecules, cell types, and cellular functions as possible. Knowledge graphs (KGs) provide such a perspective through graphically structured representations of the functional connections between biological entities. However, linking KGs of disease processes with experimental or clinical data requires their curation in a large-scale, multi-level layout. The resulting heterogeneity leads to new challenges in KG curation, data integration, and analysis. Existing approaches for small-scale applications must be adapted or combined into multi-scale tools to analyze multi-omics data in KGs. This short review reflects upon the large-scale KG approach to studying disease processes. We do not review all modeling approaches but focus on a personal perspective on.