CAREER: Novel Algorithms for Dynamic Network Analysis in Computational Biology

更广泛的意义和重要性。蛋白质是生命的主要大分子。因此，了解蛋白质在细胞中的功能是至关重要的。基因组序列研究已经彻底改变了对细胞功能的理解。然而，正如在后基因组时代所认识到的那样，基因（蛋白质）不是孤立地起作用的。相反，它们通过相互作用来完成细胞过程。这正是生物网络的模型。与基因组序列数据不同，生物网络数据能够研究复杂的细胞过程，这些过程来自蛋白质的集体行为。因此，生物网络研究有望为生命、进化、疾病和治疗原理提供新的见解。然而，目前的网络研究处理的是生物数据的静态表示，即使细胞功能是动态的。部分原因是由于生物技术在数据收集方面的局限性，无法获得实验衍生的动态生物网络数据。与静态生物网络研究相比，需要有效的计算策略来推断和分析动态生物网络，以促进对细胞功能的理解。这正是这个项目的重点。动态生物网络研究具有社会重要性的生物学应用，例如研究疾病进展，药物治疗或年龄的细胞变化，这将作为本项目的一部分进行探索。因此，该项目可促进全球健康。它也可能影响其他领域，例如社交网络。此外，该项目将产生与其研究相关的教育活动，例如通过新课程开发活动形成跨学科科学家，或通过研究监督，职业指导和面向K-12和（本科）研究生的社区外展来加强计算机科学人口，重点是女性。技术描述。这将为动态生物网络研究开辟新的计算方向。通过将静态网络拓扑与其他数据类型（如不同时间的基因表达或蛋白质丰度的测量）相结合，将开发用于推断动态生物过程背后的系统级生物网络的新算法。然后，分析动态网络数据的新方法将被开发出来，以深入了解潜在的细胞变化。例如，在静态生物网络研究中已经很好地建立的石墨烯（小子图）的思想将被带到下一个层次，以允许基于石墨烯的动态生物网络分析。此外，将设计新的计算策略以允许动态网络集群。所提出的方法将用于包括代表性动态生物学过程的协作应用：早期癌症检测和化疗耐药性，无论是在胰腺癌的背景下，还是在研究人类衰老方面。这些跨学科的应用将作为具体的模型系统来创新基础计算研究。由于网络研究跨越许多领域，实现新方法的开源软件将提供给来自不同学科的研究人员。该软件还将作为一种教育工具。进一步推动科研与教育的结合。通过新颖的网络研究课程为学生提供跨学科的培训。识字教育的目的在于提高学生的交际能力。经过验证的教学策略将用于提高学生的学习。研究监督和职业指导将提供给K-12和（本科）研究生，重点是少数民族和妇女，从而将多样性融入到项目中。跨学科研究和教育合作将使所提出的想法和结果得到广泛传播。研究结果还将通过在著名的国际会议上组织指导和讲习班来传播。

Broader significance and importance. Proteins are major macromolecules of life. Thus, understanding how proteins function in the cell is critical. Genomic sequence research has revolutionized understanding of cellular functioning. However, as recognized in the post-genomic era, genes (proteins) do not function in isolation. Instead, they carry out cellular processes by interacting with each other. This is exactly what biological networks model. Unlike genomic sequence data, biological network data enable the study of complex cellular processes that emerge from the collective behavior of the proteins. Thus, biological network research is promising to give new insights into principles of life, evolution, disease, and therapeutics. However, current network research deals with static representations of biological data, even though cellular functioning is dynamic. This is in part due to unavailability of experimentally-derived dynamic biological network data, owing to limitations of biotechnologies for data collection. Efficient computational strategies for both inference and analysis of dynamic biological networks are needed to advance understanding of cellular functioning compared to static biological network research. This is exactly the focus of this project. Dynamic biological network research has biological applications of societal importance, such as studying cellular changes with disease progression, drug treatment, or age, which will be explored as a part of this project. Thus, the project could contribute to global health. It may impact other domains as well, e.g., social networks. Also, this project will result in educational activities that are intertwined with its research, such as forming interdisciplinary scientists via novel curriculum development activities, or strengthening the computer science population via research supervision, career mentoring, and community outreach to K-12 and (under)graduate students, focusing on women.Technical description. This proposal will result in new computational directions for dynamic biological network research. New algorithms will be developed for inference of systems-level biological networks underlying a dynamic biological process, by combining the static network topology with other data types, such as measurements of gene expression or protein abundance at different times. Then, novel methods for analyzing the dynamic network data will be developed to gain insights into the underlying cellular changes. For example, the idea of graphlets (small subgraphs), which has been well established in static biological network research, will be taken to the next level to allow for graphlet-based analyses of dynamic biological networks. Also, novel computational strategies will be designed to allow for dynamic network clustering. The proposed methods will be used in collaborative applications that encompass representative dynamic biological processes: early cancer detection and chemotherapy resistance, both in the context of pancreatic cancer, as well as studying human aging. These interdisciplinary applications will be used as concrete model systems to innovate fundamental computational research. Because network research spans many domains, open-source software implementing the new methods will be offered to researchers from diverse disciplines. The software will also serve as an educational tool. Integration of research and education will be promoted even further. Interdisciplinary student training will be offered via novel courses on network research. A literate approach to education will aim to advance students' communication skills. Proven pedagogical strategies will be used to improve student learning. Research supervision and career mentoring will be offered to K-12 and (under)graduate students, with focus on minorities and women, thus integrating diversity into the project. Interdisciplinary research and educational collaborations will allow for wide distribution of the proposed ideas and results. The results will also be disseminated through tutorial and workshop organization at renowned international conferences.