Network approaches for identification of human genetic disease genes

Abstract

The identification of genes causing a genetic disease is still an important issue in the biomedical field because the list of disease genes is still incomplete while it determines the early diagnosis and treatment of fatal genetic diseases such as autism, cancer, drug resistance, and secondary hypertension. Genes associated with a particular disease or similar diseases tend to reside in the same region in a biological network and their location on the network can be predicted. Many network analysis methods have been proposed to solve this problem so far. This review first helps readers access and master the basic concepts of biological networks, disease genes, and their properties. Then, the main content is devoted to the analysis and evaluation of analytical methods recently used to find disease genes on two networks: protein-protein interaction (PPI) and cellular signaling network (CSN). We reported typical problems of identification of primary genes that cause genetic diseases and modern techniques that were widely used for solving those problems. For each technique, we also represented key algorithms so that the audience can exactly implement them for their experiments. In particular, we evaluated the performance of these algorithms in prediction of disease genes and suggested the context for their usage. Finally, the implications of the methods are discussed and some future research directions are proposed. Taken together, disease genes can often be identified from network data by two approaches: network-based methods and machine learning-based methods, and the network-based approach showed better performance in most cases because it works well even if the data size is small.