Skip to main content
. 2018 Aug 30;9:3522. doi: 10.1038/s41467-018-05624-4

Fig. 1.

Fig. 1

Overview. a Overview of the study design. Bayesian sparse linear mixed modelling (BSLMM) was used to compute SNP weights for 53 biomarkers from the ARIC study. These weights were used to compute genetically predicted biomarkers in the EHR data set and phenome wide scanning (PheWAS) was used to identify clinical phenotypes associated with the genetically predicted biomarker. b Circos plot showing the 116 significant associations (Bonferroni p < 0.05) between the genetic predictors of the ARIC biomarkers and pheWAS phenotypes. Associations are denoted by lines. Coloring is used to highlight similar groups of biomarkers and pheWAS phenotypes