Figure 1: An illustration of computation for random effects {r1;l=1,...,p } in Model 1, using a hypothetical gene set with 3 genes (variables) and 12 samples (observations). Covariance Matrix = estimated gene-gene covariance matrix . Under “Eigenvalues of the Covariance Matrix”, = 0.09802458 is the estimated first eigenvalue of . Under “Eigenvectors”, Prin 1 = is the estimated first eigenvector of . r1 is computed as a scaled product of the first eigenvalue and eigenvector, or , note that they vary according to genes, so the random effects have sub-index i in Model 1.