Applying genome-wide association study to prioritise compounds in untargeted metabolomics for Mendelian randomisation studies

Untargeted metabolomics provides an unprecedented opportunity for epidemiological studies to identify novel chemical compounds that play a role in complex disorders. Studying the causal role of compounds, however, remains challenging. Use of Mendelian randomisation (MR), a well-established approach to address causality in observational studies, depends on strong genetic determinants of the metabolites of interest. The genome-wide association study (GWAS) has so far been unsuccessful in identifying genetic factors since the computational burden of studying millions of SNPs in relation to thousands of metabolites is huge.

I aim to adapt a novel method used primarily in brain imaging, to metabolomics data and perform GWAS on thousands of features identified by various assays of untargeted NMR and mass spectrometry (MS) metabolomics. I will use data from three large epidemiologic studies, MESA, The Rotterdam Study and Airwave (n=10,000). The analysis will highlight metabolic features that are strongly regulated by genetic variants and will identify genetic variants that could be used as instrumental variables in MR analysis.

The results of this project will allow me to perform MR studies on a selected set of metabolic features in relation to various complex disorders in my future research.