Skip to main content
Fig. 3 | BMC Medicine

Fig. 3

From: Treatment- and population-specific genetic risk factors for anti-drug antibodies against interferon-beta: a GWAS

Fig. 3

Prediction of nADA. Treatment-specific prediction of the presence of nADA in the replication data. Eight PRS, the top single GWAS variant, and the top HLA allele from the discovery stage were analyzed, with sex, age, treatment preparation and duration, titration site, and ancestry components as covariates. The plots show the area under the receiver operating characteristic curve (AUC) and its 95% confidence interval (CI) calculated using bootstrapping. Bonferroni = significant after Bonferroni correction for multiple testing; nominal = nominally significant (p < 0.05); n.s. = not significant. a, b AUC of all ten prediction models for a IFNβ-1a s.c. and b IFNβ-1b s.c.; Bonferroni correction for ten tests. c The model with the highest AUC for each treatment preparation, Bonferroni correction for 160 tests (α = 3.13 × 10−4). Boxes show the prediction groups (replication data) and columns within each box the training data groups (discovery data)

Back to article page