Image-derived Spatiotemporal Models of Cellular Organization and Perturbation



 DESCRIPTION (provided by applicant): This research is aimed at developing and testing methods that will enable automated, information--‐rich studies of the subcellular organization of proteins and other biomolecules. We propose to develop the computational foundation for deciphering the spatiotemporal morphological pathways of different subcellular structures during important subcellular processes automatically from microscopy images. We will leverage our progress in the previous project period, in which we established the Cell Organizer system for building image-‐derived models of cell organization, to build new modeling capabilities for cisternal and reticular structures, for capturing the relationships between different cellular components, and for learning how aspects of cell organization change during processes such as the cell cycle. Given the importance of the endoplasmic reticulum in numerous cellular processes (as well as diseases including hematological disorders, cranio--‐ lenticulo--‐sutural dysplasia, chylomicro retention disease, etc.), we expect the modeling approaches we propose to provide important mechanistic clues as to the "normal" (non-- athological) morphological pathways of these structures and how they may be perturbed in disease. To provide valuable images to drive the development of modeling capabilities, and to simultaneously identify candidate lead compounds for treatment of diseases with ER defects, we will conduct extensive compound screening studies on cells expressing normal and mutant alleles of atlastin (which plays a critical role in ER structure). The computational methods we develop will be integrated into our open source CellOrganizer system (http://CellOrganizer.org). In addition to the applications mentioned above, we anticipate the computational frameworks we develop will be valuable for a wide range of imaging assays including drug screens, RNAi screens, cytology, and pathology.