Thanks to the many co-authors and TOPMed contributors (including @Riudecanyenc, @deirdre_tobias, and @AlisaManningPhD), and I'd love to hear any feedback and suggestions for future work!.

We report: some genetic discovery, some thoughts on statistical power grounded in realistic estimates of diet-related effect sizes and measurement error, and a model for integrating best practices from genetic epidemiology, interaction analysis, and nutritional epidemiology.

Our gene-macronutrient interaction manuscript is out in Diabetes! We set out to find instances of genetic modification of the relationship between dietary macronutrients and glycemia.

RT @JBCole150: Can genetic heritability act as an unbiased metric to compare different phenotype processing methods? Check out our (@kewes….

RT @Hyunkyungkim_: Thrilled to share that our manuscript on T2D genetic clustering is now published!.Soft clustering analysis of type 2 dia….

RT @TIMatASPPH: We had some very good in person @RallyForNIH meetings with @RepKClark and @RepLynch offices as well as virtually with @SenM….

But in seriousness, I hope this work encourages discussion about statistical power for gene-diet (and generally, GxE) interactions. Our power calcs here depend substantially on parameter assumptions; what interaction effect sizes are (1) realistic, and (2) practically relevant?.

I won't go full tweetorial while on vacation, but check out our new preprint! If you like carbohydrates and/or sobering power calculations, this one's for you!.

Shout-out to co-authors @JBCole150, @AlisaManningPhD, @miriam_udler, and others. You can browse and download our catalog of vQTLs and GxEs on the @AMP_CMDKP ( -- let us know what you think!.

2) We additionally tested for GxEs across all main-effect loci (not just vQTLs), showing that loci with vQTLs were much more enriched for underlying GxEs than standard GWAS loci. This finding supports the original motivation for this project and agrees with existing literature.

1) Because our exposome-wide approach tests many highly-correlated exposures, some proportion of identified GxEs might be redundant. For each locus, we iteratively conditioned on the most significant exposure(s), finding 132 "independent" GxEs of the 847 total.

I've previously shared the main story for the preprint: Two notable updates in the published version based on really helpful feedback from reviewers:.

In brief: prioritizing genetic loci that associate with the variance of metabolic biomarkers allowed us to test for gene-environment interactions across an "exposome-wide" set of >2000 exposures in the UK Biobank. We found 182 locus-biomarker pairs supporting 847 GxEs.

Our manuscript on variance-quantitative trait loci and exposome-wide interaction testing is now published!

RT @TamarHaspel: What should you eat?. For human health?.For planetary health?.For animal welfare?. Food is all trade-offs, all the time. M….

Great study describing a crossover trial of 4-week low/med/high spice consumption: Does anyone have thoughts on the U-shaped spice-cytokine association, with the highest level of spice consumption being generally *worse* than medium? Relevant in practice?.

RT @whsource: Interesting article relevant to "healthy user bias". h/t @LeahSamberg.

Does anybody else find this amusing? ⁦@WSJ⁩

Check out our catalog of vQTLs and GxEs here: Big thanks to co-authors @JBCole150, @AlisaManningPhD, @miriam_udler, and others, as well as the @AMP_CMDKP for hosting our browser!.

Notable examples include the interaction of triglyceride-associated variants with distinct body mass- versus body fat-related exposures as well as genotype-specific associations between alcohol consumption and liver stress at the ADH1B gene.