Abstract
The study of metabolomics and disease has enabled the discovery of new risk factors, diagnostic markers, and drug targets. For neurological and psychiatric phenotypes, the cerebrospinal fluid (CSF) is of particular importance. However, the CSF metabolome is difficult to study on a large scale due to the relative complexity of the procedure needed to collect the fluid. Here, we present a metabolome-wide association study (MWAS), which uses genetic and metabolomic data to impute metabolites into large samples with genome-wide association summary statistics. We conduct a metabolome-wide, genome-wide association analysis with 338 CSF metabolites, identifying 16 genotype-metabolite associations (metabolite quantitative trait loci, or mQTLs). We then build prediction models for all available CSF metabolites and test for associations with 27 neurological and psychiatric phenotypes, identifying 19 significant CSF metabolite-phenotype associations. Our results demonstrate the feasibility of MWAS to study omic data in scarce sample types.
Highlights
Stats were used for both the CSF metabolites and the neurological phenotypes to allow for easier replication of our results by other groups who can use our GWAS summary statistics for the CSF metabolites
The weighted median approach[120,121] is similar to Egger regression in that it helps to address pleiotropy when using multiple SNP instruments and is robust so long as no more than 50% of the instruments are pleiotropic
The same phenotype GWAS that was used in BADGERS was used for the MR analysis; otherwise, a similar phenotype from a different study was used (Supplementary Data 1, Supplementary Table 12)[26,37,91,117,118,119]
Summary
Four effects were significant after multiple testing correction, and all four of these effects were in the same direction as predicted by BADGERS but of smaller magnitude: N-delta-acetylornithine, ethylmalonate, and N6-methyllysine with schizophrenia and N-deltaacetylornithine with cognitive performance (Supplementary Data 1, Supplementary Tables 11–12; Supplementary Fig. 29). These significant effects were all for models with only one or two SNPs used as instruments
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have