MR-Base: an online resource for Mendelian randomization using summary data

The aim of our research is to develop an online tool called MR-Base, www.mrbase.org, to facilitate the application of Mendelian randomization using summary data. The objectives are: 1) to collate and harmonize summary association data from a) genome-wide and b) phenome-wide association studies of all traits into a central database; and 2) to automate the Mendelian randomization analysis pipeline. MR-Base will greatly increase the accessibility of Mendelian randomization for a wide body of researchers. Mendelian randomization is an increasingly important tool for appraising causality in observational epidemiology and can be used to prioritise intervention targets for disease prevention. However, the technique requires specialist knowledge and access to very large genetic studies, which makes the approach difficult to implement for most researchers. By systematically collating and harmonizing summary association data into a single database and by automating the analysis pipeline, MR-Base will greatly increase the scope and efficiency of Mendelian randomization for identifying targets for disease prevention. We will test the association of every genetic marker against all traits (e.g. diseases and biomarkers) in UK Biobank (where sample sizes allow). The results from these analyses will be deposited into MR-Base for Mendelian randomization (a technique that uses genetic information to predict the association between exposures and disease). We will also generate correlation statistics amongst all traits in UK Biobank, to be used for comparison with results based on genetic analyses. Using the MR-Base web app, www.mrbase.org, researchers will be able to conduct Mendelian randomization analyses of their exposures and diseases of interest. We require the full cohort as well as all diseases, risk factors and biomarkers.