Fig. 1: Power results in simulations.
From: Genome-wide discovery for biomarkers using quantile regression at biobank scale
a Homogeneous model, normal error distribution; b homogeneous model, Cauchy error distribution; c location-scale model; d local model; e model with additive and dominance effects. Power is shown when beta varies from 0.01 to 0.2. For the model with additive and dominance effects, the dominance effect is fixed at 0.2, while the additive effect varies from 0.01 to 0.2. Figures on the left show densities of the raw trait values by genotype group for one replicate. Figures in the middle show the empirical conditional quantile functions for τ = 0.1−0.9 for the same replicate for beta = 0.1 (betaA = 0.1 and betaD = 0.2 for the additive + dominance model). Statistical significance was determined by the QR rank score test described in the Methods section. Shown in black is also the LR line fitted to the data.
