TOWARD DISCOVERING DISEASE-SPECIFIC GENE NETWORKS FROM ONLINE LITERATURE

Abstract

Many human diseases are the result of abnormal interactions between multiple genes instead of single gene mutations. Discovering the interactions between these genes and their relationships to human diseases is critical for understanding mechanisms of diseases and helping design effective therapies. Valuable experimental evidence from years of industrious research by biologists can be used to help establish the underlying network of gene interactions related to human diseases. Fortunately, these information are habitually published in research journals whose abstracts are stored in a centralized, easily accessible public database called MEDLINE. To take advantage of this valuable resource, we have developed DiseasePathweaver―a computer-aided knowledge discovery system for extracting, summarizing and visualizing disease-specific gene interaction networks. Using DiseasePathweaver, a biologist can obtain a global overview of the gene interaction network related to a specific human disease, together with well-documented evidences linking to each gene and its putative interactions. We compared the gene networks of two complex human CNS diseases extracted by DiseasePathweaver to the corresponding networks from the human-curated KEGG database and found that our system can accurately cover 79% to 69% of the corresponding disease gene networks, showing the usefulness of DiseasePathweaver as a user-friendly knowledge discovery system for biologists to discover and understand gene interaction networks in complex human diseases. Free access to DiseasePathweaver is available for academic and non-profit users through http://pathweaver.i2r.a-star.edu.sg.