Fig. 1From: A simulation study investigating power estimates in phenome-wide association studiesSimulation workflow. We designed a simulation approach to test different testing parameters for a PheWAS analysis and their effect on the power estimates. For each combination of testing parameters, we generated 1000 simulated datasets. Then, we performed association testing using logistic and linear regression for binary and quantitative phenotypes respectively. The power estimates were calculated for each combination of parameter setting at significance level of α = 0.00025 (Binary trait) and α = 0.004 (Quantitative trait)Back to article page