1. An atlas of genetic influences on human blood metabolites Nature Genetics 2014 Jun;46(6)

2. Introduction discovery of mutations in severe congenital metabolic disorders, or inborn errors of metabolism early understand how genes control biochemical reactions and metabolic pathways in the human body technological advances in metabolomics and genetics a spectrum of genetic variation in human metabolism and that the loci involved often influence multifactorial traits and complex diseases, so ­called geneticallyinfluenced metabotypes (GIMs)

3. Introduction But little known about their system­ interconnectivity between genes and metabolites be translated into medical practice A comprehensive blueprint of human metabolic pathways and the genes is required.

7. Method Primary genome-wide association analysis for metabolites primary association testing was carried out at each SNP for all 486 metabolite linear regression models (assuming an additive genetic model) were used age and sex: included as covariates associations: performed by Merlin software

8. Meta-analysis Association results in TwinsUK and KORA combined using inverse variance meta-analysis based on effect size estimates and standard errors, adjusting for genomic control. Heterogeneity in each association between data sets was tested using Cochran's Q test All analyses: carried out using Metal software Method

9. Genome-wide analysis of metabolite/metabolite ratios second discovery step: carried out by testing genome-wide associations on all 98,346 (=444 × 443/2) pairwise ratios of metabolite concentrations present in both cohorts after quality control steps SNPs with low minor allele frequency (<1%) and info (<0.4):removed from analysis Method

10. Genome-wide analysis of metabolite/metabolite ratios systematically evaluated evidence for the genes within 1­Mb windows centered on the sentinel SNPs – identify function of the gene matched the relevant metabolite – Identify plausible or established biochemical link – calculate metabolite ratio Method

11. Construct a network view of genetic- metabolic interactions combine genetic and metabolite information from two cohorts connect metabolites with metabolites using Gaussian graphical models (GGMs), which are based on partial correlation coefficients connect metabolites with genetic loci on the basis of our primary GWAS results, with one link for each genome­ wide significant association verify the stability of partial correlation values in both cohorts – perform a bootstrapping­ based subsampling approach.

13. Allelic architecture of metabolic loci Refinement of independence at metabolite loci refine identity of single metabolite– and metabolite ratio–associated loci plot regional association: identify possible independent signals within each locus or to detect possible signals spanning two consecutive loci

14. Total heritability estimates genetic and environmental influences on the concentrations of metabolites present in TwinsUK after quality control steps: inferred under the ACE model (which models trait variance as a function of additive genetics, common environment and unique environment and/or error effects) Allelic architecture of metabolic loci

15. Total heritability estimates Narrow-sense heritabilities: inferred from the proportion of the total variance explained by estimated additive genetic effect estimate parameters of the ACE model: applied under multivariate normality assumptions using OpenMx software Allelic architecture of metabolic loci

16. Known heritability estimates under additive models estimated by multiple regression analysis including all associated SNPs under additive genetic models, after adjusting for covariates (age and sex in KORA; age, sex and experimental batch effect in TwinsUK) Only unrelated twin individuals :used for the analysis to avoid biases due to familial correlation Known heritability: defined as the ratio of total variance in the metabolite to the variance explained by the multiple regression models including all SNPs significantly associated with the metabolite Allelic architecture of metabolic loci

17. Epistatic interactions in known heritability estimates Frequent statistical interaction (epistasis) between genetic variants :proposed to inflate heritability estimates by creating so-called phantom heritability test for epistasis: focused on all pairs of SNPs associated with the same metabolite at genome-wide significance SNP-SNP interaction: defined as a departure from additive marginal effects and was verified by an ANOVA F test comparing the following two models with and without an interaction term, γ × SNP 1 × SNP 2 : Allelic architecture of metabolic loci

21. Pharmacological relevance Drug associations list of drugs approved by the FDA and/or the EMA was retrieved from the ChEMBL database map genes to drugs conducted using the Citeline Pharmaprojects Pipeline (see URLs; accessed 1 July 2013) to annotate targets for drugs in other stages of development. – for each metabolic locus, the pipeline was searched using the gene symbol for the metabolic locus as the search criterion in the “biological target” field