Fig. 5From: Learning mixed graphical models with separate sparsity parameters and stability-based model selectionCounts of GO terms with uncorrected p < .05 for groups of genes with expression variables linked to each discrete clinical variable in a MGM networks at different values of λ and b qp-graphs at different values of q. Edge thresholds for qp-graphs were chosen to select similar numbers of connected genes to an MGM network with λ = .1Back to article page