Network analysis by topological properties and computational modeling (Identifying critical locations and dynamic behaviors in biological and epidemiological networks)
MetadataShow full item record
Network structures were emerged broadly and frequently in real world systems ranging from the Internet, community networks to gene regulation and scientific collaborations. Several topological metrics were introduced to identify critical elements or to measure importance of elements in network systems. Existing metrics for characterizing networks have broad specificity and lack the selectivity for many applications. The purpose of this research is to introduce effective ways to identify critical elements in various networks, such as biological networks, virus spread networks and bone microstructural network, with our metrics and computational models. Real world networks are highly susceptible to disruption but robust to loss structural integrity upon targeted deletion of critical nodes. The components identified by our metrics and models are shown to be located on important paths between modules. Furthermore, these identified nodes and edges are shown to be effectively used for module finding, disease spread estimation, and bone strength analysis. Therefore, our computational modeling with measurement is an effective approach identifying important elements with unique properties that may aid in network analysis in many areas like systems biology, drug target identification, network maintenance and national security applications. Quantitative characterization of the topological characteristics of protein-protein interaction (PPI) networks can enable the elucidation of biological functional modules. As a prerequisite study, we introduce methods to identify false positive interactions and bridging cut algorithm to detect important nodes and edges for analyzing PPI structures. Since about 50% of these high-throughput data has been found to be false positives in PPI networks, results based on those datasets would not be reliable. These false positives could lead many devoted researches to erroneous biological conclusions. So, screening false positives from reliable interactions should be an indispensable work for more effective analysis on PPI data. Moreover, bridging cut algorithm enables us to find effective sub-modules in networks. The results obtained by our methods would be evaluated by comparing with experimental data of biological complexes and functions. Many real world interactions and relations are not just static but dynamic as each element and connection in various networks are evolving and changing by time. Relations in a community network changes when people create other relations and interactions in a biological network also changes during transition from one steady state to other steady states. By those evolution processes, their network structures retain the ability to change internal structures by removal unnecessary connections and replacement with newly formed relations and connections called evolving dynamics. Evolving dynamics is a fundamental property of many real world phenomena that permits adaptation to their changing environment. To understand dynamical phenomena in the real world, we design computational models to capture how a set of interactions with a common cause are spatially and temporally correlated, and how one interaction may trigger another via network influence. Then, we identify critical parameters for dynamics and how to correlate relevant interactions of each parameter. As interesting applications, we introduce two computational network models, Avian Influenza disease spread and control model, and microstructural bone network model. In the first model, we focus on numerical analyses on experimentation results with a computational model of mitigating disease spread in spatial networks. We describe the model system and the details of our calculations, and present the results of this study. Then, we briefly discuss the properties of fixed-radius random networks considered here followed by details of calculations on the spatial and temporal features of our computational model of disease spread process and containment process. We also analyze both the damage of the disease spread process and the damage of the containment process with parameter estimations. At last, we show the spatial features of the model using a real world geological network as a possible application. For the microstructural bone network model, we focus on analyzing bone strength and bone remodeling dynamics. Since the conventional measurement, bone mineral density (BMD), of bone strength is not the sole factor responsible for the fracture risk, we introduce a new approach to improve bone strength estimation and fracture risk assessment. In addition, bone is a dynamic, living tissue whose structure and shape continuously adjusts to mainly provide structural framework. In this respect, we propose a computational model of bone microstructure and dynamics which is capable of quantitative assessment of bone mineral density and bone remodeling dynamics. In this application, we generate a bone network based on our bone model reflecting bone microstructure and then introduce a mathematical model of remodeling dynamics among osteoblast and osteoclast. It allows us to calculate cell population dynamics and changes in bone mass at multiple sites of bone remodeling. Last, we analyze bone quality and identify critical elements in bone networks.