Extended Data Fig. 3: Examples of prioritization of putative risk factor-protein-disease causal triangles.
From: Mapping biological influences on the human plasma proteome beyond the genome
a-c, Examples of potential molecular mediators of causal associations between risk factors and diseases. We report effect estimates from Mendelian Randomization (beta coefficients for risk factor to proteins, and OR for risk factor or proteins to diseases. aCausal effects based on the Wald ratio estimate reported by Pietzner et al. bCausal effect reported by Gaziano et al. in individuals with eGFR ≤ 60 mL/min/1.73m2. We note that in c, we present the inverse beta coefficient as we do this for reduced eGFR, rather than eGFR (as performed in the analysis) to orient the triangle to direction of increased risk. d, Alternative scenario representing independent effects converging on the same protein. This example might be best explained by the known modifying effect of the lead cis-pQTL for ADH1B (rs1229984) on alcohol consumption. Therefore, ADH1B might be independently associated with the risk of gout and ALT through alcohol consumption; and in turn, the observed MR finding between ALT and ADH1B plasma levels might rather reflect reverse causality. Created with BioRender.com.
