Fig. 3
From: Incorporating functional annotation with bilevel continuous shrinkage for polygenic risk prediction
Evaluation of AUC for UK Biobank analysis results. Left panel: KEGG functional annotations were used for the analysis of the proposed; Right panel: Refseq gene-based functional annotations from ANNOVAR were used for the analysis of PRSbils. From top to bottom: type II diabetes, BMI, LDL. For type II diabetes, summary statistics were obtained result from a GWAS analysis of 407,701 white British UK Biobank participants. For BMI, we used the GWAS results from GIANT Consortium with 332,153 participants with European ancestry. For LDL, GWAS results from the GLGC Consortium with 188,578 participants with European ancestry were used