An atlas of polygenic risk score (PRS) associations across the human metabolome

Please select a PRS:

Please select a metabolite:



This web application can be used to query findings from a systematic analysis of 129 polygenic risk scores and 249 circulating metabolits using high-throughput nuclear magnetic resonance data from the UK Biobank study1,2. We encourage users of this resource to conduct follow-up analyses of associations to investigate potential causal and non-causal metabolic biomarkers. Age-stratified results can be used to investigate how potential sources of collider bias (e.g. statin therapy) may influence findings in the full sample


To query results from our atlas please select parameters on the left hand side of this page before clicking the 'Search Atlas' button. This will generate results in the Tables of Results tab.

P-value threshold - Please select whether you wish to query results for using polygenic risk scores derived using SNPs based on P<0.05 or P<5x10-08.

There are 2 ways to query the results of our atlas depending on the Input Type selected:

A single PRS against 249 metabolites - Select a polygenic risk score from the drop-down menu and click the 'Search Atlas' button. This will query the findings between your choice of polygenic risk score with all 249 circulating metabolites from our analysis. Forest plots3 will be generated to display results with metabolite traits grouped and coloured by their subcategories.

All PRS on a single metabolite - Select a circulating metabolite from the drop-down menu and click the 'Search Atlas' button. This will query the associations between all polygenic risk scores in our atlas with your chosen trait. Forest plots will be generated to illustrate these results where risk scores are grouped and coloured based on their subcategories.

The Download button can be used to download a comma-separate value (csv) file for the results you have queried.


An atlas of associations between polygenic risk scores from across the human phenome and circulating metabolic biomarkers by Si Fang, Michael V Holmes, Tom R Gaunt, George Davey Smith & Tom G Richardson. medRxiv (2021) doi:


For any queries please contact: