Empirical Validation of GA4GH's Data Use Ontology with NIH Datasets' Data Use Limitations

项目摘要 该项目的目标是验证全球基因组学与健康联盟（GA 4GH）数据使用本体 (DUO)使用超过700个NIH数据集来告知其在扩展数据共享中的使用。DUO的开发旨在提供 用于描述基因组数据允许的二次使用的标准词汇。通过标准化数据使用 在全球范围内，DUO加速数据共享，符合参与者的同意以及道德和法律的 规定作为世界上最大的生物医学研究公共资助者，NIH的基因型数据库和 表型（dbGaP）拥有超过7，582个数据集，截至3月，通过dbGaP提供超过330，000个变量 2021年;使dbGaP中的NIH资助的数据集成为DUO验证的理想测试案例。在 在这个项目中，我们将与NIH数据访问委员会（DAC）合作，调整其数据集的现有数据使用 数据限制（DUL）使用DUO中的术语。因此，我们将确定哪些DUL和DUO术语 是否平等以及两者之间存在差距，并讨论是否以及如何与DAC解决差距。我们 我的目标是在最初的DUO之后，在正式出版物中总结这些关键的发现和建议 在布罗德研究所验证120个数据集，并通过与GA 4GH数据使用和研究的通信 身份工作流领导（劳森先生是其中的一员），以供他们在发展 朵此后，我们将与NIH DAC进行讨论，以确保所有NIH DUL都能映射到 DUO中的机器可读术语。一旦通过对DUO、NIH DUL、 或两者兼而有之，我们将在布罗德研究所的数据使用监督中使用适当的DUO术语标记所有数据集 系统（DUOS）数据集目录。DUOS数据集目录将升级为查询数据集， 允许的数据使用（例如，DULs），以便研究人员能够过滤掉可能不会被批准的数据集 访问。DUOS的决策支持算法将访问请求中的DUO术语与DUO匹配 dbGaP数据集上的术语。DUOS将使用它来自动分类请求，并提出决定， 是否授予访问权限，可以很容易地由DAC审查。DUL输入经过精确校准， 在DUOS中启用的功能，NIH DAC和科学政策办公室将能够很好地编纂 他们的DAR审查政策到DUOS的决策支持算法。这将使自动决策， 针对典型请求的DAR，并显著降低DAC的负担和DAR周转时间， 研究人员

Project Summary The goal of this project is to validate the Global Alliance for Genomics and Health (GA4GH) Data Use Ontology (DUO) using >700 NIH datasets to inform its use in scaling data sharing. The DUO was developed to provide a standard vocabulary for describing permitted secondary uses of genomic data. By standardizing data use language globally, the DUO expedites data sharing compliant with participants’ consent and ethical and legal regulations. As the largest public funder of biomedical research in the world NIH’s database of Genotypes and Phenotypes (dbGaP) holds over 7,582 datasets on over 330,000 variables available via dbGaP as of March 2021; making the NIH-funded datasets housed in dbGaP an ideal test case for the validation of the DUO. In this project we will work with NIH data access committees (DACs) to align their datasets’ existing data use limitations (DULs) with data use terms from the DUO. As a result, we will identify which DULs and DUO terms have parity and where gaps exist between the two, and discuss if and how to address the gaps with DACs. We aim to summarize these key findings and recommendations in a formal publication, following on the initial DUO validation of 120 datasets at the Broad Institute, and via a communication to GA4GH Data Use & Research Identities work stream leadership (of which Mr Lawson is a member) for their consideration in evolving the DUO. Thereafter, we will advance discussions with the NIH DACs to ensure all NIH DULs can be mapped to machine-readable term(s) in the DUO. Once completed through either adjustments to the DUO, the NIH DULs, or both, we will tag all datasets with the appropriate DUO terms in the Broad Institute’s Data Use Oversight System (DUOS) Dataset Catalog. The DUOS Dataset Catalog will be upgraded to query datasets’ by their permitted data use (e.g., DULs) so that researchers are able filter out datasets they likely will not be approved to access. The DUOS’ decision-support algorithm will match the DUO terms on access requests to the DUO terms on the dbGaP datasets. DUOS will use this to automatically triage requests and propose a decision to grant access or not that can easily be reviewed by DACs. With the DUL inputs accurately calibrated and their functionality enabled in DUOS, the NIH DACs and Office of Science Policy would be well-positioned to codify their DAR review policy into DUOS’ decision support algorithm. This would enable automated decisions on DARs for typical requests, and significantly reduce the burden on DACs and DAR turnaround time to researchers.