Integrative Visualization of Spatiotemporal Tumor Atlases

点击翻译按钮获取中文摘要

Project Summary With the rise of new high-throughput technologies that enable the measurement of biomolecules at the single-cell level within the spatial tissue context, emerging human tumor atlases promise to illuminate the role of cellular interactions in cancer. Spatiotemporal tumor atlases integrate molecular, cellular, and structural information, as well as clinical data. Interpretation of the resulting 2D and 3D maps will lead to new insights into molecular processes that drive tumorigenesis and eventually guide the development of new cancer treatment strategies. The development of specialized visual exploration tools to support this discovery process is critical. However, these tools must be designed to address several challenges: a large number of features such as cells or molecules that need to be visualized, the need to link across diverse modalities such as genomic and imaging data, the presence of multiple timepoints and organizational scales, as well as the volume of data generated by many assays. Furthermore, the cancer research community includes a broad spectrum of user audiences with varying data visualization needs. Therefore, we propose to create a framework for integrative, web-based visualization of human tumor atlases. Our work will be guided by user-centered design processes and we will be collaborating with users in the cancer research community to elicit user needs and to evaluate our visualization tools. This will allow us to design and implement visualization tools that are appropriate for the targeted user audiences. Given the diversity and size of the datasets generated by assays used to build tumor atlases, we will design this framework to be extensible and scalable from the ground up. We are proposing three distinct aims for this project. One aim is to develop a modular, web-based toolkit for visual analysis of multimodal spatial single-cell cancer datasets. This toolkit will also be available through R and Python APIs for integration into computational notebooks. This will enable data analysts and software developers to connect any visualizations created with the toolkit to state of the art computational analysis techniques, which complement the visual exploration supported by the visualizations. Another aim of our work is the design and implementation of novel methods for comparative and longitudinal visualization of spatiotemporal tumor atlas data sets. Finally, we will also build a web-based platform that will allow any cancer researcher to create their own interactive visualizations of multimodal spatiotemporal tumor atlas datasets through a graphical user interface. Users will also be able to share links to these visualizations with other researchers and the general public.