Fig. 4

Illustration of internal working pattern of DMM. X-axis shows 500 samples and y-axis shows the phenotype. For each figure, there are 4 sub-figures. The first one shows how the prediction by DMM (orange) fits the true phenotype (yellow). The second shows how the fixed-effect estimation component (blue) fits the phenotype (yellow). The third one shows the how the confounding factor correction component (red) fits the phenotype (yellow), and the fourth one shows how the confounding factor correction component (red) fits the confounding effects (green). (a) and (b) are two sets of visualizations of the simulation experiments with two different random seeds