Fig. 2: Power comparison for simulations with varying proportions of true causal SNPs pcausal = (0.001, 0.01, 0.05, 0.1), true expression heritability \({{{ h}}}_{{{ e}}}^{{{2}}}=\left({{0.1,0.2,0.5}}\right)\), and phenotype heritability \({{{h}}}_{{{ p}}}^{{{2}}}\in\left({{0.05,0.875}}\right)\). | Nature Communications

Fig. 2: Power comparison for simulations with varying proportions of true causal SNPs pcausal = (0.001, 0.01, 0.05, 0.1), true expression heritability \({{{ h}}}_{{{ e}}}^{{{2}}}=\left({{0.1,0.2,0.5}}\right)\), and phenotype heritability \({{{h}}}_{{{ p}}}^{{{2}}}\in\left({{0.05,0.875}}\right)\).

From: SR-TWAS: leveraging multiple reference panels to improve transcriptome-wide association study power by ensemble machine learning

Fig. 2

Expression was simulated 1000 times per scenario. Similar patterns were observed as shown in Fig. 1, where Avg-valid+SR models obtained the highest test R2 for gene expression imputation across all 11 of 12 scenarios and SR-TWAS models performed best in the scenario with \({p}_{{\mbox{causal}}}=0.01\) and \({h}_{e}^{2}=0.5\). Power is calculated as the percent of simulation iterations which have TWAS p value \( < 2.5\times {10}^{6}\). Color and shape indicate model: blue/square = PrediXcan-GTEx, gold/up-triangle = TIGAR-ROSMAP, green/down-triangle = TIGAR-ROSMAP-valid, black/cross = Naive, red/circle = SR-TWAS, purple/diamond = Avg-valid+SR.

Back to article page