The extended Chow-Liu algorithm was developed for undirected graph construction. The algorithm depends on a penalized likelihood criterion for the addition of edges. A major advantage of this method is the ability to accommodate both discrete and continuous variables, simultaneously. The method is also well suited for high-dimensional data. Through the modification of weights, the algorithm searches for a minimal BIC (Bayesian Information Criterion) forest. In this project, we examine this method using data from a breast cancer study comparing women with and without a mutation in P53 tumor suppression gene. The dataset has 250 observations including 58 cases with p53 mutation and 195 controls without the mutation, and 1000 gene expression features. The gRapHD package is leveraged for implementation. Results reveal a hub gene, CENPA, and its neighbors, which are known to play an important role in breast cancer. We applied forward stepwise search using minimal BIC forest as a start model and compared the degree distribution for variables with more than four direct connections, which consistent results in terms of the structure of the graphical model.