Investigating the mechanisms underlying disease using multiOmics data

Advances in genotyping technologies and computational capacities have brought new insights into the genetic mechanisms influencing disease. Using genetic markers (SNPs) and health and disease information from many thousands of individuals, Genome-Wide Association Studies (GWAS) have uncovered genetic differences between individuals that cause disease. We also know that epigenetic mechanisms - changes in a chromosome other than DNA sequence changes - play a role in determining disease through regulation of gene activity and expression. DNA methylation is a type of epigenetic modification, sometimes under nuclear genetic control.
Generation Scotland (GS) is a cohort of 20,000 individuals with information on many health-related traits including body measures, indicators of cardiovascular and metabolic health and electronic health records. GS participants have information on over 500,000 SNPs, and 5,000 participants have DNA methylation information at around 700,000 locations ('sites') in the genome. Through this DNA methylation dataset that is large, in numbers of individuals and numbers of sites surveyed, compared to published studies, we have an unprecedented insight into the genetics underlying DNA methylation and have discovered that SNPs that control DNA methylation at sites linked to obesity also affect obesity itself. We will now have DNA methylation data on 5000 more individuals from the same cohort.
We will use this large scale DNA methylation data, metabolite and protein measures and publicly available GWAS and expression data to get new insights on the genetic and epigenetic mechanisms that control disease. Our project will have 3 main stages:
1. We will use the newly available DNA methylation information in GS and combine it with our previous study to further uncover SNPs that affect methylation levels across individuals (mQTLs). We are interested both in mQTLs that control methylation at nearby sites (in 'cis') and more distantly (in 'trans'), as they provide different information on how methylation is controlled.
2. We will use the cis-mQTLs from 1 to elucidate the mechanism through which SNPs (mQTLs) affect methylation and disease. For instance: does the SNP affect DNA methylation and, through methylation, the phenotype? Or do SNPs affect the phenotype and that, in turn, changes DNA methylation? Or are both DNA methylation and phenotype affected by the same SNP (or nearby SNPs) but independently of each other? We will answer those questions using Mendelian Randomisation (MR) and reverse MR, our own data and publicly available GWAS results for disease-related phenotypes obtained from very large populations such as UK BioBank. These very large studies allow the detection of SNP-phenotype associations that smaller studies miss, and we will have information on SNP-phenotype associations for over 700 phenotypes of medical relevance.
3. To help answer the questions in 2, we will also use publicly available data on gene expression in various tissues (GTEx data), sequence data and proteomics and metabolomics gathered in populations managed in-house at the IGMM.
This project tackles active research areas and offers the opportunity to develop new computational and statistical approaches, in particular incorporating new Machine Learning methods.