Flexible & Executable Provenance in Data-Intensive Biomedical Research: A Flexible Research Data Service

 描述（由申请人提供）：追踪科学数据和断言的谱系对于“曾经确保科学保真度的制衡”至关重要（柯林斯和塔巴克，2014年）。随着数据变得普遍数字化，自动和大规模地生成和跟踪谱系增加了结论的有用性和质量。数据密集型科学的一个重大挑战是生成科学信息的血统来源，同时促进检索和重新执行。我们假设这种能力提高了断言的可重复性，并使数据对社会更有用。我们的目标是为出处，数据完整性，存储和可复制的工作流程构建应用程序编程接口，使研究人员能够记录，检索和重新运行科学谱系。拟议研究的理由是，通过能够回顾性地重现结果并了解其来源以供未来使用，科学数据的价值得到了提高。出处也便于测量数据的重要性--它的影响。在强有力的前期工作的指导下，我们将通过追求两个具体目标来测试我们的假设：（1）为出处，数据管理，数据完整性和可重复执行的工作流构建API，（2）提供一个用于存储和部署容器化计算环境的平台，该平台也可以作为可重复数据科学的学习实验室。这种方法在专注于灵活性和适应生物医学科学的无数用例方面具有创新性，同时为培训可重复数据科学的研究人员提供了一个中心。通过创建一个开源的“灵活的研究数据服务”，拟议的研究将大大影响我们的能力，使我们在生物医学研究的投资更有用。

 DESCRIPTION (provided by applicant): Tracing the lineage of scientific data and assertions is critical for the "checks and balances that once ensured scientific fidelity" (Collins & Tabak, 2014). As data become pervasively digitized, generating and following lineages automatically and at scale increases the usefulness and quality of conclusions. A significant challenge in data-intensive science is generating that lineage-the provenance-of scientific information, while facilitating retrieval and re-execution. We hypothesize such capabilities improve the reproducibility of assertions and make data more useful to society. Our objective is to build application programming interfaces for provenance, data-integrity, storage, and reproducible workflows that empower researchers to record, retrieve, and re-run scientific lineages. The rationale for the proposed research is that the value of the scientific data is enhanced by being able to retrospectively reproduce a result and by understanding its origins for future use. Provenance also facilitates measurement of data's importance-its impact. Guided by strong preliminary work, we will test our hypothesis by pursuing two specific aims: (1) Building APIs for provenance, data management, data integrity, and re-executable workflows, (2) Providing a platform for storing and deploying containerized compute environments that also serves as a learning laboratory for reproducible data science. This approach is innovative in focusing on flexibility and accommodating the myriad use cases across biomedical science, while pro- viding a hub for training investigators in reproducible data science. By creating an open source "Flexible Re- search Data Service", the proposed research will significantly impact our ability to make our investments in biomedical research more useful.