DataJoint Pipelines for Neurophysiology

神经科学领域，特别是BRAIN Initiative，正在积累大量用于管理、分析和共享数据的信息学资源。然而，每个实验室都必须解决类似的问题，将这些工具集成到他们实验的特定工作流程中：数据输入，采集，处理，分析和可视化。我们创建了DataJoint：一个用于实验数据管道和协作自动化工作流程的开源通用框架。无论是独立还是在我们公司的支持下，数十个研究小组都采用了DataJoint用于各种数据模式，如多电极电生理学，钙成像，光学和电子显微镜，行为跟踪，感觉刺激，转录组学和光遗传学。在这些复杂的数据生态系统中，DataJoint使科学家团队能够在清晰的工作流程中进行协作，同时持续共享数据并保持数据完整性。 为了将共享数据管道传播和集成到神经科学实践中，我们建议编译和分发一个用于神经生理学的DataJoint Pipelines库。我们将从基于DataJoint的开放项目中提取基本的主题，并通过数据模态将它们组织成一系列精心策划的相互兼容的模块，将数据收集和计算连接在一起。这些模块已记录在案，并准备重新部署，将配备数据输入辅助工具（电子实验室书籍），根据现有项目改编的交互式图形用户界面（例如，集成用于图像分割和尖峰分选的流行算法的可定制数据处理步骤），以及现有脑图谱、坐标框架、用于细胞分类的本体的集成，以及病毒和转基因株系的命名法。为了促进统一的本体论，我们将使这些管道与新兴的神经数据标准相协调。这项工作的结果将作为一个在线资源进行管理，该资源将沿着展示管道以及示例数据、可执行演示和部署指南。参观者将能够定制，扩展和联合收割机组合成适合他们特定实验的完整管道，并学习构建新解决方案的技术。这些模块将支持各种计算架构，包括专用硬件和云服务。我们还将开展基于网络的培训和现场黑客马拉松，以帮助加速这些工具在神经科学实验室中的采用。Vathes LLC在执行这项工作方面具有独特的优势，因为我们为IARPA的MICrONS，International Brain Lab和BRAIN Initiative Ul9 pro等合作项目共同开发数据管道。该项目的成果将使许多新的研究小组能够采用DataJoint，并将信息学工具有效地整合到神经科学实践中。

The field of neuroscience and the BRAIN Initiative in particular are amassing a wealth of informatics resources for managing, analyzing, and sharing data. Yet every lab must solve similar sets of problems to integrate these tools into the particular workflows of their experiments: data entry, acquisition, processing, analysis, and vi­sualization. We created DataJoint: an open-source general framework for experiment data pipelines and col­laborative automated workflows . Either independently or with support from our company, dozens of research groups have adopted DataJoint for diverse data modalities such as multielectrode electrophysiology, calcium imaging, light and electron microscopy, behavior tracking, sensory stimulation, transcriptomics, and optoge­netics. In these complex data ecosystems, DataJoint enables groups of scientists to collaborate with clear work­ flows while sharing data continuously and conserving data integrity. To disseminate and integrate shared data pipelines into neuroscience practice, we propose to compile and dis­seminate a library of DataJoint Pipelines for Neurophysiology. We will extract essential motifs from DataJoint­ based open projects and organize them by data modality into a collection of curated, mutually-compatible modules stringing together data collection and computation. Documented and ready for de novo deployment, the modules will feature ancillary tools for data entry (electronic lab books), interactive graphical user inter­ faces adapted from existing projects (e.g. customizable data processing steps integrating popular algorithms for image segmentation and spike sorting), and integration of existing brain atlases, coordinate frameworks, ontologies for cell classification , and nomenclatures for viruses and transgenic lines. To foster uniform ontolo­gies, we will harmonize these pipelines with emerging neurodata standards. The results of this work will be administered as an online resource showcasing the pipelines along with sample data, executable demos, and guidelines for deployment. Visitors will be able to customize, extend, and combine the modules into complete pipelines suitable for their particular experiments and learn the techniques for building new solutions. The modules will support diverse computing architectures, including specialized hardware and cloud services. We will also conduct web-based trainings and in-person hackathons to help accelerate the adoption of these tools in working neuroscience labs. Vathes LLC is uniquely positioned to perform this work as we co-develop data pipelines for collaborative projects such as IARPA's MICrONS, International Brain Lab, and BRAIN Initiative Ul9 pro jects. The output of this project will enable many new research groups to adopt DataJoint and to effectively integrate informatics tools into neuroscience practice.