W42. RARE VARIANT EFFECTS IN METABOLITES ASSOCIATED WITH MAJOR DEPRESSION

Previous studies have observed altered metabolic profiles in individuals with depression. A recent study by Davyson et al. (2023) used common genetic variants as instruments in a Mendelian randomisation design, to test for causal associations between 249 metabolomic measures in the UK Biobank and Major Depressive Disorder (MDD). They found that docosahexaenoic acid, along with omega-3 to omega-6 ratios, were causally associated with depression. In addition, genetic variation in the FADS region was implicated in this relationship. A limitation of this study is that the high linkage disequilibrium between common variants in the FADS region makes it difficult to identify the causal genes. Our study uses recently available short-read whole-genome sequencing (srWGS) datasets to test for associations between rare genetic variants and both depression and the causally linked metabolites from Davyson et al. Rare variants (i.e., those variants that occur in < 1% of the population) are more likely to be pathogenic and less affected by LD and are therefore a valuable tool for identifying causal genes in gene-dense regions. Our primary analysis used a burden testing approach, in which rare variants within a specific region are collapsed into a single predictor and tested for associations with a phenotype. This helps overcome the reduced power to detect effects, that occurs when variants appear in only a small fraction of the population. Restricting our analysis to variants with a minor allele frequency below 1%, we conducted burden testing between the major genes in the FADS region (FADS1, FADS2, FADS3, TMEM258, FEN1 and MYRF) and MDD. We conducted analysis seperately in the UK Biobank (n = 488,377) and the All of Us cohort (n = 296,405). We conducted both European-focussed and mixed ancestry analysis. Subsequently, we tested for association between rare variants in the FADS region and 5 metabolite measures suggested by Davyson et al.: Omega-6, Omega-3 to total fatty acids percentage, Omega-3 to Omega-6 ratio, DHA, and Unsaturation. This analysis was conducted in UKB only (n = 273,477) as metabolites were not available in All of Us. Additionally, we conducted leave-one-out (LOVO) variant analysis in order to identify individual associated with MDD and metabolites. Using burden tests, we found no significant associations between rare-variant burden and MDD in the FADS region. However, we found multiple significant associations across the region between rare variants and all 5 metabolite measures. LOVO analysis identified a number of individual rare variants that were significantly associated with metabolite levels. In particular, high-impact frame-shift variants were identified in FADS1 and FADS2 which were significantly associated with variation in all 5 metabolite levels. In addition, these frame-shift variants only affected specific transcripts, implicating interactions between rare variants and splicing variation in molecular phenotypes. Our initial analysis demonstrates a significant impact of rare variants in metabolites causally associated with MDD. Follow-up analysis is underway to test for association between the prioritised variants and MDD. In addition, we are testing for the impacts of covariates such as BMI and deprivation. Once completed, we plan to expand our analysis to other metabolites.