A biologically informed framework for instrument selection in dietary Mendelian randomization using chemosensory genetics

Abstract Background Mendelian randomization (MR) is increasingly used to strengthen causal inference in nutritional epidemiology. However, dietary MR studies often rely on instruments selected from genome-wide association studies of self-reported intake based solely on statistical significance, increasing vulnerability to pleiotropy and reverse causation and potentially violating key MR assumptions. We aimed to develop and evaluate a biologically informed framework for selecting valid genetic instruments for dietary exposures, leveraging genes encoding taste and olfactory receptors that influence chemosensory perception and shape food preferences and dietary behaviour. Methods We prioritised 1,214 nonsynonymous variants (minor allele frequency ≥ 1%) across 30 taste and 295 olfactory receptor genes. Associations with 140 food-liking traits were tested in UK Biobank participants aged 37 to 73 years. Candidate variants were evaluated using a multi-stage filtering pipeline comprising replication in an independent younger cohort (Avon Longitudinal Study of Parents and Children, age 25), concordance between food liking and intake, exclusion of associations with socioeconomic status, assessment of food specificity accounting for linkage disequilibrium and co-consumption, and directionality testing to reduce reverse causation. Retained variants were used as instruments in MR analyses of cardiometabolic outcomes. Results We identified 268 variants within 101 olfactory and 16 taste receptor genes associated with 96 food-liking traits. Filtering yielded 24 candidate instruments for 20 foods. The instrument for onion liking satisfied all pre-defined criteria for classification as high confidence. As an illustrative MR application, genetically proxied onion liking was associated with lower systolic and diastolic blood pressure and reduced risk of type 2 diabetes, with no evidence of effects on body mass index, glycaemic traits, or serum lipid levels. Conclusions Instrument selection guided by chemosensory genetics provides a scalable strategy for dietary MR that can improve instrument credibility and reduce susceptibility to pleiotropy and reverse causation. However, this approach prioritises biological specificity at the cost of fewer available instruments. This framework supports more robust causal evaluation of diet–disease relationships and strengthens inference in nutritional epidemiology and public health research.