Using Mendelian Randomization to identify modifiable environmental exposures and molecular pathways that causally affect disease risk

Rationale
Mendelian Randomization (MR) is a statistical methodology that uses genetic variants to test whether observational associations between environmental risk factors and disease represent causal relationships (Davey-Smith & Hemani, 2013). To date, a limitation of the approach has been low statistical power, in that large sample sizes are required to demonstrate causality. The UK Biobank Study is a large cross sectional study that has measured genotypic and phenotypic information on 100,000 - 500,000 individuals and so represents an opportunity to perform MR in a powerful sample of individuals. The aim is to utilize the UK BioBank and other large scale population based resources to investigate causal relationships between modifiable exposures and disease using MR. It addition the relationship between disease and thousands of allelic scores that proxy molecular phenotypes will be investigated in an attempt to data mine possible causal relationships in the data (Evans et al. 2013).

Aims & Objectives
(1) To use the UK Biobank resource to perform Mendelian Randomization of modifiable environmental exposures and disease/disease related traits in a very large sample of individuals.
(2) To identify causal relationships between thousands of molecular phenotypes (i.e. gene expression, gene methylation) and disease/disease-related traits using a data-mining approach in a very large sample of individuals (Evans et al. 2013).

Methods
The project will involve performing MR on several medically relevant observational relationships that have previously not able to be addressed because of small sample size. Allelic scores that proxy thousands of molecular phenotypes including gene transcription, methylation and levels of metabolites will then be constructed. The student will then screen for association between these instruments and phenotypes within the UK10K Biobank and other large scale population based resources and follow up interesting associations with more formal MR analyses.