Fig. 3: Validation of rvPRS models. | Communications Biology

Fig. 3: Validation of rvPRS models.

From: Studying rare variant polygenic risk scores using whole exome sequencing and imputed genotype data

Fig. 3

Only rvPRS models that showed significant associations and risk reclassification in the validation dataset are displayed. A The left solid bar plot presents Nagelkerke’s R2 or adjusted R2 while the right forest plot displays the perSD OR or Beta depending on the trait class. B This forest plot represents the NRI and IRI values comparing the “Full model” with the “Null model” and the “rvPRS+cvPRS model” with the “cvPRS model”. The error bars represent the 95% confidence intervals (CIs). For 12 of these 13 traits (all except BRE), the final selected rvPRSs were derived from single-SNP association summaries in the IMP data. Final selections: AF, IMP (INFO > 0.8) Maf4; AST, IMP (INFO > 0.3) Maf1; CA, IMP (INFO > 0.8) Maf1; DM2, IMP (INFO > 0.3) Maf4; HER, IMP (INFO > 0.3) Maf3; HT, IMP (INFO > 0.3) Maf2; IH, IMP (INFO > 0.3) Maf4; BRE, WES; BMI, IMP (INFO > 0.3) Maf1; SBP, IMP (INFO > 0.8) Maf1; DBP, IMP (INFO > 0.3) Maf1; RHR, IMP (INFO > 0.3) Maf1; FEV, IMP (INFO > 0.3) Maf1.

Back to article page