arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1803.07433cs.SE
收藏

数据来源跟踪作为生物医学虚拟研究环境的基础

Data provenance tracking as the basis for a biomedical virtual research environment

Richard McClatchey

摘要:

在复杂的数据分析中,越来越重要的是,除了长期保存数据集之外,还应获取关于数据集使用情况的信息,以确保结果的可重复性,核实他人的工作,并确保数据用于具体分析的适当条件。基于科学工作流的研究开始意识到捕获数据的这种来源以及用于处理、转换和对这些数据进行研究的活动的好处。支持工作流程的开发及其在(协作)生物医学分析中的使用的一种方法是通过使用虚拟研究环境。然而,网格/云计算的动态和分布式特性使得来源信息的捕获和处理成为一个主要的研究挑战。此外,大多数工作流起源管理服务仅针对面向数据流的工作流而设计,研究人员现在意识到单独或单独跟踪数据或工作流不足以支持科学过程。协作研究所需的是在一个完整的编排虚拟研究环境(VRE)中可追溯和可再现的出处支持,使研究人员能够根据所使用的数据集和流程定义他们的研究,监控和可视化他们的分析结果,并记录他们的结果,以便其他用户可以调用所获得的知识来支持后续研究。我们扩大了在neuGRID和N4 U项目中开展的工作,提供了一个所谓的虚拟实验室,为通用的VRE奠定了基础,其中包括生物医学数据集(图像,实验室测试结果,患者记录,流行病学分析等)。用于处理这些数据的工作流程(流水线),以及它们的来源数据和结果集都被CRISTAL软件捕获。

英文摘要:

In complex data analyses it is increasingly important to capture information about the usage of data sets in addition to their preservation over time to ensure reproducibility of results, to verify the work of others and to ensure appropriate conditions data have been used for specific analyses. Scientific workflow based studies are beginning to realize the benefit of capturing this provenance of data and the activities used to process, transform and carry out studies on those data. One way to support the development of workflows and their use in (collaborative) biomedical analyses is through the use of a Virtual Research Environment. The dynamic and distributed nature of Grid/Cloud computing, however, makes the capture and processing of provenance information a major research challenge. Furthermore most workflow provenance management services are designed only for data-flow oriented workflows and researchers are now realising that tracking data or workflows alone or separately is insufficient to support the scientific process. What is required for collaborative research is traceable and reproducible provenance support in a full orchestrated Virtual Research Environment (VRE) that enables researchers to define their studies in terms of the datasets and processes used, to monitor and visualize the outcome of their analyses and to log their results so that others users can call upon that acquired knowledge to support subsequent studies. We have extended the work carried out in the neuGRID and N4U projects in providing a so-called Virtual Laboratory to provide the foundation for a generic VRE in which sets of biomedical data (images, laboratory test results, patient records, epidemiological analyses etc.) and the workflows (pipelines) used to process those data, together with their provenance data and results sets are captured in the CRISTAL software.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑