Abbreviations

• Chr = Chromosome
• Pos = Position
• EA = Effect allele
• OA = Other allele (non-effect allele)
• EAF = Effect allele frequency
• EAF SE = EAF Standard Error
• EAF Min = EAF Minimum
• EAF Max = EAF Maximum
• MAC = Minor allele count
• SE = Standard error
• Het. = Heterogeneity
• Het. DF = Het. degrees of freedom

Notes on study design

A total of 14,307 European ancestry individuals contribute to the analysis (5,841 EPIC-Norfolk samples and 8,466 INTERVAL samples). The mean age in the discovery sample was 59.8 years in EPIC Norfolk and 44.0 years in INTERVAL with 53.3% females in EPIC Norfolk and 48.8% in INTERVAL.
Plasma metabolites were measured using the DiscoveryHD4® platform (Metabolon, Inc., Durham, USA). After QC, 913 metabolites were present in both INTERVAL and EPIC-Norfolk. For each metabolite, the values were median normalised for run day (median set to 1 for run-day batch). Normalised values were winsorized to 5 s.d. from the mean. Each value was then log (natural) transformed prior to calculating the residuals adjusted for covariates such as age, gender, batch, study center and ancestry principal components. Prior to the genetic analysis, these residuals were standardised to a mean of 0 and s.d. of 1.
Study participants were genotyped using the Affymetrix UK Biobank Axiom array and imputed using a combined 1000 Genomes Phase 3-UK10K panel (INTERVAL, EPIC-Norfolk) and the HRC rel. 1 panel (EPIC-Norfolk). Variants with imputation quality INFO < 0.4 or MAC of <= 2 were excluded. Genetic association analyses were performed within two studies separately using BOLT-LMM or SNPTEST. Study specific results were then pooled using inverse variance weighted fixed effect meta-analyses for up to 17 million genetic variants.

Please refer to the methods section of our publication for further details.