Skip to main content
. 2017 May 11;33(18):2882–2889. doi: 10.1093/bioinformatics/btx314

Fig. 3.

Fig. 3.

Performance evaluation in presence of irrelevant studies. Left panels: Performance of risk variant identification measured by AUC. Right panels: Performance of risk prediction measured by correlation between the observed phenotype values and the predicted values. In all panels, {X,y} in the x-axis indicates the performance of individual-level data only, q corresponds to the performance of integrating the individual-level data with P-values from a study simulated using parameter uk=Pr(γj(k)=1|γj=1){0.05,0.1,0.3,0.5,0.7,0.9,1}, and s indicates the stepwise performance achieved at the s-step