Boss: A cloud-based data archive for electron microscopy and x-ray microtomography

摘要 科学上丰富的高分辨率神经成像量的产生在程度上继续增加， 由于新的电子显微镜（EM）和X射线显微断层扫描（XRM）成像系统的进步， 项目和数据处理方法。随着这些技术在全球范围内的普及， 神经科学界，并持续投资于新的能力开发和研究计划- 根据BRAIN计划，当前和持续需要社区数据存储来支持 未来的数据生成，以及使二级科学，以帮助证明广泛的数据收集和 这些数据的处理成本，并传播新的科学工作流程和发现。 我们的BossDB数据生态系统目前作为BRAIN Initiative存档，用于高分辨率EM和XRM 数据，为多个BRAIN存储近1 PB的公共图像、分割和连接组学数据 项目和更广泛的神经科学社区。虽然BossDB生态系统继续成功地 支持多种研究努力，因为我们进入了大脑编程的新阶段和新一代 连接组学研究的10- 100倍更大的千万亿次图像体积收集，其他功能包括 需要确保有效和高性能的数据存储、处理、注释和访问， 方案和社区需要。因此，在R24-MH 114785的拟议更新中，我们将开发几个 BossDB的性能和可扩展性改进：提高数据的效率和成本效益 存储，支持社区工具和标准的集成，用于大数据处理和注释 缩放神经影像，并提供优化的查询服务，使其更容易进行二次分析 托管的数据集。继续支持这一关键的社区资源将使世界有可能 利用数PB的免费、公开可用和公平（可查找、可验证、可互操作和可重用）数据 在BossDB中，神经科学界已经依赖于连接组学研究。 BossDB系统将继续通过灵活的流程进行开发，以支持社区的投入 利益相关者，包括来自多个正在进行的专注于数据和实验标准的BRAIN项目的利益相关者， 以及多模态数据生态系统整合。与目前的BossDB系统一样，我们将继续整合 用于数据存储、访问和可视化的社区工具，重点关注用于数据处理的新工具集成- 执行、注释和查询。考虑到数据收集可能需要更多的分布式数据存储， 我们还将开发更强大的数据导出和本地云混合同步 技术. BossDB是一个经过专业设计的社区资源，它支持许多实验 在连接组学和神经科学研究中，本文提出的发展将进一步使更多 更广泛用户的科学影响。

ABSTRACT The generation of scientifically rich, high resolution neuroimaging volumes continues to increase in extent and rate due to the advancement of new Electron Microscopy (EM) and X-ray Microtomography (XRM) imaging sys- tems and data processing methodologies. With the increased availability of these technologies throughout the neuroscience community, and with sustained investment in new capability development and research program- ming from the BRAIN initiative, there is a current and continued need for community data storage to support future data generation, as well as to enable secondary science to help justify the extensive data collection and processing costs for these data and to propagate new scientific workflows and discoveries. Our BossDB data ecosystem currently serves as the BRAIN Initiative archive for high-resolution EM and XRM data, storing nearly one petabyte of public image, segmentation, and connectomics data for multiple BRAIN programs and for the broader neuroscience community. While the BossDB ecosystem continues to successfully support multiple research endeavors, as we enter into a new phase of BRAIN programming and a new generation of connectomics research with 10-100x larger petascale image volume collections, additional capabilities are required to ensure efficient and performant data storage, processing, annotation, and access for anticipated programmatic and community needs. Thus, in this proposed renewal of R24-MH114785 we will develop several performance and scalability improvements to BossDB: improve the efficiency and cost-effectiveness of data storage, support the integration of community tools and standards for data processing and annotation of large scale neuroimagery, and provide an optimized query service to make it easier to conduct secondary analyses of hosted datasets. Continued support of this critical community resource will make it possible for the world to leverage petabytes of free, publicly available, and FAIR (Findable, Accessible, Interoperable, and Reusable) data in BossDB that the neuroscience community already depends on for connectomics research. The BossDB system will continue to be developed through an agile process to enable inputs from community stakeholders, including from multiple ongoing BRAIN programs focused on data and experimental standards, and multi-modal data ecosystem integration. As with the current BossDB system, we will continue to integrate community tools for data storage, access, and visualization, with a focus on new tool integration for data process- ing, annotation, and querying. Given the likelihood for more distributed data storage needs as data collections increase significantly in scale, we will also develop more robust data export and local-cloud hybrid syncing technologies. BossDB is a professionally-engineered community resource that has enabled many experiments in connectomics and neuroscience research, and the developments proposed herein will further enable more scientific impact by a broader range of users.