Learning Latent Variable Gaussian Graphical Model for Biomolecular Network with Low Sample Complexity

Figure 4

Topological for the biomolecular network of medulloblastomas. The sizes of nodes correspond to the degrees of their interactions. The colors represent similar classes according to the pairwise Pearson correlations of gene expressions. Demonstrated here is the absolute essentiality of using conditional independence to interpret the apparent correlations between gene expressions. (a) represents the WNT subgroup; (b) the SHH subgroup; (c) the subgroup 3; and (d) the subgroup 4. The regularization parameters are chosen to optimize BIC score.

(a)

(b)

(c)

(d)