Biologically informed instrument selection for dietary Mendelian randomization using chemosensory receptor variants

Area of Science:

• Nutritional Epidemiology
• Genetic Epidemiology
• Causal Inference

Background:

• Mendelian randomization (MR) is crucial for causal inference in nutritional epidemiology.
• Traditional dietary MR studies face challenges like pleiotropy and reverse causation due to reliance on self-reported intake data.
• A biologically informed framework is needed to select valid genetic instruments for dietary exposures.

Purpose of the Study:

• To develop and evaluate a novel framework for selecting valid genetic instruments for dietary exposures.
• The framework utilizes genes encoding taste and olfactory receptors, which influence food preferences and behavior.
• To improve the reliability of causal inference in dietary research.

Main Methods:

• Prioritized nonsynonymous variants in taste and olfactory receptor genes (MAF ≥1%).
• Tested associations with food-liking traits in UK Biobank and validated in a younger cohort.
• Employed a multi-stage filtering pipeline including replication, concordance, socioeconomic status exclusion, specificity assessment, and directionality testing.

Main Results:

• Identified 268 variants in 117 chemosensory receptor genes associated with 96 food-liking traits.
• Filtered down to 28 candidate instruments for 24 foods, with onion liking being high confidence.
• Genetically proxied onion liking showed associations with lower blood pressure and reduced type 2 diabetes risk.

Conclusions:

• A biologically informed strategy using chemosensory receptor genes enhances genetic instrument selection for dietary MR.
• This approach mitigates issues of pleiotropy and reverse causation, leading to more robust causal evaluations.
• The framework strengthens inference in nutritional epidemiology and public health research.