Developing an integrated system for biological network exploration

Network analysis and visualization have been used in systems biology to extract biological insight from complex datasets. Many existing network analysis tools either focus on visualization but have limited scalability, or focus on analysis but have limited visualizations. The separation of analyzing the raw data from visualizing the analysis results causes systems biologists to jump between forming a question, building a massive network, identifying a subnetwork for visualization, and using the visualization as feedback and inspiration for the next question. This iterative process can take several days, making it difficult for researchers to maintain the mental map of the questions queried. In addition, biological data is stored in different formats and has differing annotations, thus systems biologists often run into hurdles when merging large or heterogeneous networks. The polymorphic nature of the datasets presents a challenge for researchers to integrate data to answer biological questions. A more systematic method for merging networks, resolving data conflicts, and analyzing networks may improve the efficiency and scalability of heterogeneous multi-network analysis. Towards improving and pushing forward multi-network analysis to help a researcher easily combine multiple heterogeneous biological data networks to answer biological questions, this dissertation reports several accomplishments that provide (i) a set of standard multi-network operations, (ii) standard merging rules for heterogeneous networks, (iii) standard methods to reproduce network analyses, (iv) a single integrated software environment that allows users to visualize and explore the network analysis results and (v) several examples applying these methods in biological analysis. These efforts have culminated in three academic publications.