iSubgraph: Integrative Genomics for Subgroup Discovery in Hepatocellular Carcinoma using Graph Mining and Mixture Models

iSubgraph is an unsupervised integrative framework to discover patterns of miRNA-gene networks, observed frequently up- or down-regulated in a group of patients and to use such networks for patient stratification in cancer patients.

iSubgraph (MATLAB code)

Figure 1. Schematic overview of the iSubgraph algorithm. Flow chart starts with transforming microarray data into graph representation (left); and then continues with graph mining-based method to identify significant miRNA-gene co-modules (top right); and proceeds to tumor subclassification by the mixture model (bottom right).

Figure 2. Graph Mining steps for a sample small dataset. (A) Normalized gene and miRNA expression levels in tumor and adjacent nontumor tissues. (B) Steps of correlation analysis. From left to right, the matrices show the correlation coefficients between genes and miRNAs computed from expression profiles, target predictions by analysis of mRNAs with seed-complementary sequences, and the actual targets determined by correlation analysis. (C) Template bipartite graph representing miRNA-gene interactions and their correlation types. (D) All patient graphs constructed for all patients based on template bipartite graph. Threshold for UP and DOWN tags is ±1. (E) All frequent closed subgraphs for a support threshold of 2 patients.

Figure 3. Mixture model for cancer subgroup discovery using plate notation. Circles indicate random variables. Shaded circles denote observed values. Outer rectangles indicate fixed parameters. The directed edges show dependencies between variables and parameters. Capital letters represent the size of parameters (vector or matrix) and plate repetition.