Fig. 2.

Precision-recall curves comparing performance of BTH versus three other methods and example plots of underlying discrete simulated data. Panel A: increasing mean effect size: π_maf = 0.2, n = 300, $\beta \in \{0,0.2,1\},\hspace{0.17em}\hspace{0.17em}\text{log}(\alpha )=0.1$; Panel B: increasing the variance effects: ${\pi }_{maf}=0.2$, n = 300, $\beta =0.5,\hspace{0.17em}\hspace{0.17em}\text{log}(\alpha )\in \{0,-0.1,-0.2\}$; Panel C: increasing minor allele frequency: ${\pi }_{maf}\in \{0.05,0.2,0.3\}$, n = 300, $\beta =0.5,\hspace{0.17em}\hspace{0.17em}\text{log}(\alpha )=0.1$; Panel D: increasing sample size: ${\pi }_{maf}=0.2,\hspace{0.17em}n\in \{300,500,1000\},\hspace{0.17em}\beta =0.5,\hspace{0.17em}\hspace{0.17em}\text{log}(\alpha )=0.1$