Data Science Core

数据科学核心

• 批准号: 10516499
• 负责人: Douglas H Kelley
• 金额: $ 40.58万
• 依托单位: UNIVERSITY OF ROCHESTER
• 依托单位国家: 美国
• 财政年份: 2022
• 资助国家: 美国
• 起止时间: 2022-08-01 至 2027-07-31
• 项目状态: 未结题
• 来源: https://reporter.nih.gov/project-details/10516499
• 关键词: Algorithms; Anatomy; Animals; Archives; Artificial Intelligence; Attention; Blood Volume; Brain; Brain imaging; Characteristics; Classification; Code; Collaborations; Computer software; Data; Data Files; Data Science; Data Science Core; Data Scientist; Data Set; Data Storage and Retrieval; Data Store; Electroencephalography; Goals; Human; Image; Individual; Information Retrieval; Infrastructure; Investigation; Lead; Learning; Link; Liquid substance; Magnetic Resonance Imaging; Measurement; Measures; Metadata; Methods; Modernization; Mus; Noise; Participant; Pattern; Publications; Reading; Recording of previous events; Research; Resources; Running; Sampling; Scientist; Shapes; Sleep; Software Tools; Suggestion; System; Time; Training; Uncertainty; Universities; Variant; Viral; Visualization; Work; Writing; anatomic imaging; base; cerebrospinal fluid flow; code development; computational platform; data access; data infrastructure; data sharing; data tools; doctoral student; experience; experimental study; file format; hemodynamics; image processing; improved; in vivo; innovation; instrument; large datasets; machine learning algorithm; mathematical model; member; multimodal data; multimodality; neural circuit; neuroregulation; novel; open source; organizational structure; programs; relating to nervous system; repository; searchable database; signal processing; simulation; solute; statistical and machine learning; supercomputer; synergism; tool; two-photon; user-friendly; virus core

Abstract, Data Science Core The goal of this proposal is to characterize and quantify how neural circuits control cerebrospinal fluid (CSF) flow and solute clearance during sleep and wake, in both mice and humans. Achieving that goal will require combining multimodal data from mathematical models and various experiments in both species. The Data Science Core will provide the essential infrastructure and produce innovative data methods to enable powerful synergy among the Projects. First, because the Projects will produce enormous data sets, and because the data will be multimodal (MRI, two-photon imaging, simulation results, EEG, and more), productive collaboration will require careful attention to storing, organizing, processing, analyzing, and internally sharing the data. Aim 1 is to provide data infrastructure and staffing for seamless integration of all Projects, for synergistic code development and sharing, and for rapid analysis of multimodal data via efficient workflows, always leveraging existing best practices. Second, discovering the causal links among neural control, CSF flow, and solute clearance will require tools and analyses that do not yet exist. Aim 2 is to build transformative tools leveraging quantitative, data-driven methods to measure CSF flow, its drivers, their neural control mechanisms, and the resulting efflux. Third, the novel data and tools produced in Projects 1-4 will have potential to advance the field, enabling a range of new scientific questions, especially if they are disseminated widely. Aim 3 is to document the software tools we develop, annotate the data we produce, and share both publicly. The Data Science Core will build on existing methods and tools previously developed by the co-PIs, take advantage of best practices in the field, and combine the expertise of experienced data scientists to build novel tools. Documented code will be shared via GitHub. We will use a two-tiered system for sharing data internally, in which each Project will leverage their existing data workflows and keep some data locally; meanwhile, data shared among Projects will be stored on the BlueHive supercomputer at the University of Rochester, available to all U19 participants. We will develop a clear organizational structure for shared data, storing metadata in sidecar files and implementing a searchable database to facilitate collaboration and interaction among the Projects. We will keep a searchable stock list of tools provided by the Virus Core, along with delivery times and histories. Annotated data will be shared publicly via repositories upon publication. A full-time data scientist will be employed to lead day-to-day Core activities, along with two PhD students.

摘要，数据科学核心 该提案的目标是表征和量化神经回路如何控制脑脊液 (CSF) 流动 小鼠和人类在睡眠和醒来期间的溶质清除率。实现这一目标需要结合起来 来自两个物种的数学模型和各种实验的多模态数据。数据科学核心 将提供必要的基础设施并产生创新的数据方法，以实现之间的强大协同作用 项目。首先，因为这些项目将产生巨大的数据集，并且因为这些数据将被 多模式（MRI、双光子成像、模拟结果、脑电图等），需要高效的协作 认真关注数据的存储、组织、处理、分析和内部共享。目标 1 是提供 数据基础设施和人员配置，用于所有项目的无缝集成、协同代码开发和 共享，并通过高效的工作流程快速分析多模式数据，始终利用现有的最佳技术 做法。其次，发现神经控制、脑脊液流动和溶质清除之间的因果关系需要 尚不存在的工具和分析。目标 2 是利用定量、数据驱动的方式构建变革性工具 测量脑脊液流量、其驱动因素、神经控制机制以及由此产生的流出的方法。第三， 项目 1-4 中产生的新颖数据和工具将有潜力推动该领域的发展，从而实现一系列新的 科学问题，特别是当它们被广泛传播时。目标 3 是记录我们使用的软件工具 开发、注释我们产生的数据并公开共享。数据科学核心将建立在现有的 共同PI先前开发的方法和工具，利用该领域的最佳实践，并结合 经验丰富的数据科学家的专业知识来构建新颖的工具。记录的代码将通过 GitHub 共享。 我们将使用两层系统在内部共享数据，其中每个项目将利用其现有数据 工作流程并在本地保存一些数据；同时，项目之间共享的数据将存储在BlueHive上 罗彻斯特大学的超级计算机，可供所有 U19 参与者使用。我们将制定明确的 共享数据的组织结构，将元数据存储在 sidecar 文件中并实现可搜索 数据库以促进项目之间的协作和交互。我们将保留可搜索的库存清单 Virus Core 提供的工具以及交付时间和历史记录。带注释的数据将公开共享 发布后通过存储库。将聘请一名全职数据科学家来领导日常核心活动， 和两名博士生一起。