guest ::
| Home > Publications database > Management of data and metadata - an exemplary electrophysiological workflow for collaborative data analysis |
| Home > Publications database > Management of data and metadata - an exemplary electrophysiological workflow for collaborative data analysis |
| Poster (After Call) | FZJ-2017-07593 |
; ; ;
2017
Abstract: The complexity of neuroscientific experiments and their analysis has grown to a degree where special effort andattention is required to guarantee their reproducibility. In addition, collaborations across different laboratoriesand countries are becoming a standard work setting, which increases the need for comprehensivedocumentation of data and metadata, and explicit, formal descriptions of the data analysis process. Theavailability of software tools that support scientists in the various steps of this process is therefore indispensable[Denker and Grün (2016) In: Brain-Inspired Computing: Brain Comp 2015, Springer]. Moreover, it is necessary toestablish the workflows that link the various tools, and to develop simple interfaces for them.Here we demonstrate how such a reproducible, structured, and comprehensible workflow for anelectrophysiological experiment, covering the preparation, annotation and analysis data, can be set up in acollaborative environment by the combination of multiple state-of-the-art open-source projects. The workflowfeatures the Electrophysiology Analysis Toolkit (Elephant), which represents the central analysis resourceoffering methods ranging from the analysis of ensemble spike data to population signals, such as local fieldpotentials [http://neuralensemble.org/elephant/]. Elephant is based on the generic standardized datarepresentation for electrophysiological data provided by the Neo library [Garcia et al. (2014) Front Neuroinf8:10]. In addition, Neo is able to interface with a range of data formats commonly used in electrophysiology. Theopen metadata Markup Language (odML) is used as the hierarchical structure to store metadata related toelectrophysiological experiments [Grewe et al. (2011) Front Neuroinf 5:16]. Furthermore, odMLtables extendsthe accessibility of odML by providing an interface to a tabular metadata representation, e.g., using Excel[https://github.com/INM-6/python-odmltables]. Finally, NIX is a newly developed scheme designed to combineelectrophysiological data and metadata in a single, standardized format [https://github.com/G-Node/nix], andlinks the Neo and odML data models. We discuss multiple mechanisms that allow to describe the workflow itself,and show how it may be implemented, e.g., using the Collaboratory of the Human Brain Project[https://collab.humanbrainproject.eu]. While focusing on electrophysiology, many concepts of this workflow arealso transferable to different types of experimental environments.
|
The record appears in these collections: |