An integrated network motif based approach to identify colorectal cancer related genes

Identifying colorectal cancer related genes is crucial to diagnose and treat this disease. Many conventional approaches based on gene expression data have been developed to identify disease genes through patterns of gene activity. These approaches usually ignore other heterogeneous information, for example, gene-gene interactions, epigenomic data and gene expression data. We propose a novel integrated approach based on network motif to identify colorectal cancer related genes, which considered the network topological characteristic and coordinated changed pattern characteristic of gene expression data, epigenomic data. Screened network motifs and motif´s genes are respectively used as classification features to distinguish colorectal cancer samples from normal samples. Result shows they achieve almost the same accuracy in classification as the known cancer marker genes. Functional enrichment analysis reveals that the genes of the screened motifs are annotated to important biological process. Most of them can be validated to highly associate with the development of colorectal cancer based on previous studies. We not only provide a method for identification of disease related genes but also add a new perspective to integrate heterogeneous data and mine subgraph with significant biological characteristics pattern.