This title appears in the Scientific Report :
2021
Approaches for improving rigor and efficiency in sharing complex neurophysiological data
Approaches for improving rigor and efficiency in sharing complex neurophysiological data
Sharing complex neurophysiological data, e.g. in a collaboration or in a publication, in such a way that theycan be understood and used is a challenge. Issues to be solved are to define adequate metadatarepresentations to describe the experiment, find standardized formats to store the data, or to ca...
Saved in:
Personal Name(s): | Grün, Sonja (Corresponding author) |
---|---|
Denker, Michael | |
Contributing Institute: |
Computational and Systems Neuroscience; IAS-6 Jara-Institut Brain structure-function relationships; INM-10 Computational and Systems Neuroscience; INM-6 |
Imprint: |
2021
|
Conference: | 14th Meeting of the German Neuroscience Society, Online (Online), 2021-03-22 - 2021-03-30 |
Document Type: |
Abstract |
Research Program: |
Helmholz Metadata Collaboration Human Brain Project Specific Grant Agreement 3 Neuroscientific Foundations Digitization of Neuroscience and User-Community Building Neuromorphic Computing and Network Dynamics |
Publikationsportal JuSER |
Sharing complex neurophysiological data, e.g. in a collaboration or in a publication, in such a way that theycan be understood and used is a challenge. Issues to be solved are to define adequate metadatarepresentations to describe the experiment, find standardized formats to store the data, or to capture thedetails of the data preprocessing workflows. For most of these challenges easy or generic solutions do notyet exist that could be used “out of the box”. Where solutions do exist, the knowledge and expertise on howto employ them in the context of a particular experiment is often scattered in the community. Inconsequence, the implementation of a research data management strategy in an experimental laboratory isoften a balance act between ensuring a certain level of rigor, reproducibility and documentation on the onehand, and efficiency (in terms of time and personnel) of designing and implementing strategies on the otherhand.Moreover, the results of this balancing act may turn out to be frustrating: even with comparatively highinvestment in improving the quality of data and metadata descriptions and data curation process, the resultwill often be disappointing in the sense that the benefits associated with good research data managementare not realized. This holds in particular when a lack of standards for data formats, structures, and commonvocabulary prevents the curated dataset to be easily (e.g., automatically) used by other scientists, analyzedby existing tools, or integrated into data stores. Thus, despite the efforts, the data remain difficult to use andcomprehend. Moreover, the solutions for data management developed on a project-by-project basis, whilebased on promising ideas, are rarely developed to a level of stability and generality required forstandardization, and tend to be lost after the project’s lifetime.To overcome this unsatisfactory situation, we here argue for the need to promote the level of standardizationas a community effort. To this end, we analyze the process of data acquisition and preprocessing ofelectrophysiological data [1,2]. This example includes a number of aspects that tend to complicate the datacuration process, such as a data acquisition process that spans over multiple years of a running experiment,or the need for collaboration across labs. We describe the steps that we required to accomplish the goal ofcurating data to a degree that they become sharable and publishable in sufficient detail to ensure easy andunsupervised reuse in data analysis workflows [3]. In this endeavor, we consider conceptual considerationsunderlying the design of the data acquisition and curation process and highlight the standardizationstrategies and tools (e.g., [4-7]) that can help in reducing the effort of handling the data. Some of these toolsare coordinated and harmonized as part of the EBRAINS e-infrastructure for neuroscience [8], which iscommitted to contribute and shape a layer of interoperable compute and data services for futureneuroscience. Likewise we will outline the needs where complementary standardization is still non-adequateor non-existent, and suggest how community efforts coordinated by the NFDI-Neuro consortium [9] couldaddress, seed, and drive towards concrete solutions. Getting into a habit of treating the design of datamanagement for new experiments early on in harmony with community standards and recommendations willgive scientists the opportunity to spend more time analyzing the wealth of electrophysiological data theyleverage with low-barrier collaborations, rather than dealing with data formats and data integrity. |