Deciphering the Non-Coding Genome using Millions of Diverse Whole-Genome Sequences

More than 90% of genetic variants associated with human disease lie in poorly understood parts of our genome. These regions do not directly code for proteins, and are therefore termed 'non-coding'. Instead, they regulate when, where and how much of each protein is produced by coding genes. For the first time, the complete genetic sequences, medical records, and extensive health data of over 1 million people are becoming available. We now therefore have the ability to identify functionally impactful regions in the non-coding genome for common diseases like obesity and Type 2 diabetes (T2D). However, it is extremely challenging computationally and methodologically to analyse data on 1,000,000 complete whole genomes. Interpretation is also a substantial challenge. Despite these challenges, my preliminary work demonstrates the huge potential for discovery. In 200,000 individuals in UK Biobank we identified rare non-coding variants in a switch for a gene called HMGA1 that means people are up to 5cm taller. Further, by analysing levels of 1,500 proteins in the blood, we have demonstrated that there are thousands of impactful non-coding variants in these genetic switches. Circulating protein levels provide an excellent way to test our approach to analysing the non-coding genome because they are the primary product of genes, and so directly impacted by nearby non-coding genetic variants. This project will build on my initial work by developing our whole genome sequence analysis framework to make it efficient, cost-effective and publically available. By first expanding my analysis to 3,000 circulating protein levels, I will (i) develop computational and methodological frameworks for performing efficient, low-cost whole-genome sequencing analysis. Analysis of circulating protein levels will provide insight into the most relevant annotations, most powerful statistical methods, and characterise the spectrum of non-coding variant effects on common human traits. These insights will enable me to apply my framework to BMI (body mass index) and T2D, analysing up to 1 million individuals with diverse ancestries from the largest studies in the world with whole-genome data (UK Biobank, All of Us & TOPMed). My foscus on metabolic traits aligns with Exeter as a centre for world-leading diabetes research, whilst allowing me to apply my own expertise, in order to (ii) fine-map common genetic variants to causal genes and (iii) identify novel and functionally-important non-coding elements. Discoveries I make for T2D and BMI have the potential to lead to novel drug targets, improve patient quality of life and improve precision medicine. I will release an open-source pipeline for cloud-based whole-genome analysis, and provide a template for analysis of the non-coding genome, which will provide unique insight into human disease, and guide efforts to understand the biological function of genetic variation.