CAREER: Image-Based Computational Methods for Understanding Spatiotemporal Dynamics of Cellular Processes

细胞的内部环境是高度异质和动态的，但组织得很好。在这种环境下，细胞过程的成功完成需要正确的分子或分子复合物在正确的时间、正确的地点发挥作用。因此，了解细胞过程的动态时空行为对于在系统水平上理解其分子机制至关重要。该项目的研究目标是通过开发和应用所需的计算分析和建模方法，从细胞时空动态图像中提取细胞过程的机械知识。该技术的发展将受到基本和具体的生物学问题的驱动和指导，这些问题涉及神经元细胞轴突货物运输的时空调节，以及有丝分裂纺锤体中的微管和运动易位，它们分别是一维（1D）和二维（2D）空间细胞过程的代表。具体研究目标是：1)结合单粒子跟踪和超分辨率成像技术表征细胞时空动态；2)分别利用隐马尔可夫模型和时空点过程统计来表征一维细胞过程的微观和宏观时空动力学，并结合生物物理建模来理解轴突货物运动的模式；3)通过非刚性形状配准和均匀晶格空间划分来表征二维细胞过程的时空动态，并结合生物物理建模来理解有组织的纺锤体微管和运动易位。过去半个世纪生物学的进步使鉴定细胞的所有分子成分成为可能。这些成分如何在空间和时间上发挥作用和相互作用来驱动特定的细胞过程，以及控制这些过程的一般原理是生物学的基本问题。该项目将提供在广泛的生物学研究中解决这些问题所需的计算方法和软件。这些方法和软件将免费提供给更广泛的研究界。与此同时，该项目将推进和促进卡内基梅隆大学（CMU）生物图像计算分析和理解（也称为生物图像信息学）的教学和培训，特别是对来自工程、计算机科学和生物学专业的学生，以及来自卡内基梅隆大学-匹兹堡大学计算生物学联合博士课程的学生。本项目提供的研究和教育资源将用于研究生和本科生的教学和培训，并参与CMU生物医学工程系和计算机科学学院组织的外展项目。

The inner environment of the cell is highly heterogeneous and dynamic yet exquisitely organized. Successful completion of cellular processes within this environment requires the right molecule or molecular complex to function at the right place at the right time. Understanding dynamic spatiotemporal behaviors of cellular processes is therefore essential to understanding at the systems level their molecular mechanisms. The research objective of this project is to extract mechanistic knowledge of cellular processes from images of their spatiotemporal dynamics by developing and applying required computational analysis and modeling methods. The technological development will be driven and guided by fundamental and specific biological questions regarding the spatiotemporal regulation of axonal cargo transport in neuronal cells as well as microtubule and motor translocation in mitotic spindles, which are representatives of cellular processes that are one-dimensional (1D) and two-dimensional (2D) in space, respectively. The specific research aims are: 1) To characterize spatiotemporal cell dynamics by combining single particle tracking with super-resolution imaging techniques; 2) To represent microscopic and macroscopic spatiotemporal dynamics of 1D cellular processes using hidden Markov models and spatiotemporal point process statistics, respectively, and to combine with biophysical modeling to understand patterned axonal cargo movement; and 3) To represent spatiotemporal dynamics of 2D cellular processes via non-rigid shape registration and uniform lattice space partitioning, and to combine with biophysical modeling to understand organized spindle microtubule and motor translocation.Advances in biology over the past half a century have made it possible to identify all the molecular components of a cell. How these components function and interact in space and time to drive specific cellular processes and what general principles govern these processes are fundamental questions to biology. This project will provide computational methods and software that are required to address these questions in a broad range of biological studies. The methods and software will be made freely available to the broader research community. In the meantime, the project will advance and promote teaching and training of computational analysis and understanding of biological images (also referred to as bioimage informatics) at Carnegie Mellon University (CMU), especially for students from engineering, computer science, and biology programs, and from the CMU-University of Pittsburgh joint PhD program in computational biology. Research and educational resources made possible by this project will be used for teaching and training of graduate and undergraduate students and for participating in outreach programs organized by the Department of Biomedical Engineering and the School of Computer Science at CMU.