Design And Implementation Of Integrated Cancer Information Visualization Plugin Based On Cytoscape

With the accomplishment of Human Genome Project and the fast development of genome-sequencing technologies,tremendous genome data were produced in the area of cancer research.Making full use of these data to perform data mining is very critical in cancer related research,and at the meanwhile,to present the information behind these massive data visually,we must take some appropriate visualization tools.And therefore,developing a visualization tool according to the feature of special data will make sense in exploring some of these cancer related data.In this thesis,we analyzed the visualization forms of currently most used visualization tools in visualizing cancer information,and their strengths and weaknesses as well.Then we decided to develop a visualization plugin based on Cytoscape 3 for integrated cancer information.We also described our design of the plugin and some implementation details in this thesis.The plugin was designed within the research content of pharmacogenomics with two purposes:do clustering on the interaction network of cancer related genes,then matche the gene clusters with a gene-drug-disease network and visualize the results in frames,researchers could predict new drug targets by observing the results;put all the gene,drug and disease data into a database for query.As for clustering,we chose two algorithms:MCODE and MCL,and the matched results would be visualized in frames,users could extract a subnetwork in dense region by double-clicking a node.Besides,users could query a node’s information by clicking a node.The data we used in the plugin includes gene interaction network and gene-drug-disease network.The gene interaction network comes from a gene related data file that we got from UCSC Gnome Browser,we separated the gene interaction data from the file and turned the data into an Excel file that can be import into Cytoscape directly.The gene-drug-disease network comes from DrugBank,we classified all the drugs and diseases according to the short description and handled the data blanks manually,and then we put all the data into a database for the convenience of query.The plugin in this thesis involves a new analytic method for discovering potential drug target in pharmacogenomics.This plugin displays gene clusters and their related drugs and diseases at different levels,researchers can predict new drug targets through observing matched results,then to help guiding the use of targeted drugs and exploring the mechanisms of existed targeted drugs.