EAGER: Identifying Blockmodel Functional Modules across Multiple Networks

点击翻译按钮获取中文摘要

Broader Significance and Importance: Due to the high complexity of analyzing high-throughput omics data, most of the existing computational methods separately analyze the data collected from different sources. Furthermore, they typically assume that the available prior biology knowledge, such as molecular interactions manifested as biological networks, is accurate. As the existing curated molecular interactions and functionalities across public databases still have unsatisfactory coverage or consistency, it is critical to develop effective analysis methods that can achieve biologically meaningful solutions by integrating diverse evidence from multiple biological networks. The objective of the proposed research is to develop a network-based mathematical framework and a set of new computational algorithms for the integrative analysis of multiple networks. The proposed research has strong transformative potentials in network biology. If successful, it can eventually lead to computational tools for more accurate and reliable identification of novel biomarkers and functional pathways. Beyond that, through the ongoing collaborations with biologists and physicians, it will open up new applications of network analysis methods to improve our understanding of complex human diseases. The interdisciplinary nature of this proposal promises to foster cross-fertilization of ideas between engineering and biology through research and education.Technical Description: The proposed research investigates integrative analysis of multiple biological networks, which are often noisy, to robustly identify biologically significant functional modules. The proposed mathematical framework provides a platform to address both critical issues in multiple network analysis regarding the computational complexity and biological significance by simultaneously analyzing multiple networks in a modular space. The advantage of integrative analysis of multiple biological networks is two-fold: First, cellular functional pathways that carry out critical functionalities are likely to be conserved across different organisms. Multiple network analysis will improve the performance of functional module identification. Second, new evidence from the analysis of identified modules may effectively transfer previously accrued knowledge to more confident curation and annotation of molecular relationships and the underlying cellular mechanisms. The proposed research and education activities are to: 1) design a new mathematical model for multiple biological network analysis to identify network modules for better understanding functional organization of cells and the complex cellular mechanisms; 2) devise effective and efficient optimization algorithms, including mathematical programming and stochastic optimization algorithms, to solve the optimization problems at different levels of complexity; 3) evaluate the performance of the proposed methods by constructing biologically realistic benchmark datasets; 4) apply the methods to systems biology research through collaboration with biomedical researchers; and 5) integrate research findings into the education and training of students with various academic backgrounds in the interdisciplinary field of computational network biology.