Integrating UK Biobank personal health information with a heterogeneous knowledge network to gain new insights into complex diseases
Approved Research ID: 61342
Approval date: October 26th 2020
The aim of this research is to integrate the UK Biobank data with a biomedical knowledge graph to improve our understanding of complex diseases. Our group has developed SPOKE (Scalable Precision medicine Oriented Knowledge Engine) a resource that includes biomedical information from 29 publicly available databases. By combining real-word medical data from the UK Biobank to this collection of biological knowledge, we will generate specific "signatures" (SPOKE profiles) for each patient. Each individual SPOKE profile is essentially a different version of the same knowledge graph, in which each element is assigned a different relative importance, based on the patient data. These signatures or embeddings are biologically meaningful and machine-readable, providing both insights into these phenotypes and opportunities for prediction of clinical outcomes.
We propose to first build on our previously published work by analyze the complete set of phenotypes in the UK Biobank dataset using SPOKE. This will allow us to identify novel associations between diseases, genes, medications (including opportunities for drug repurposing) and biological processes. No individual data will ever be made public or shared in any way.