guest ::
| Home > Publications database > Sharing Electrophysiological Data and Metadata on HBP Platforms – An Example Collaboratory Workflow |
| Home > Publications database > Sharing Electrophysiological Data and Metadata on HBP Platforms – An Example Collaboratory Workflow |
| Poster (After Call) | FZJ-2017-01810 |
; ; ;
2017
Abstract: IntroductionThe Human Brain Project (HBP) [1] aims at creating and operating a European scientific Research Infrastructurefor the neurosciences. A main goal is to gather, organise and disseminate data describing the brain and itsdiseases on the basis of experimental as well as simulated data. Therefore a lot of effort is put into thedevelopment of tools for data registration, storage, access and sharing. The most prominent data type availablethrough the HBP to date are anatomical data and data describing single cell dynamics. However, the need toinclude experimental and simulated large-scale functional data, and in particular, electrophysiological activitydata, has been widely recognized. Such data are primarily created in neuronal network simulations as a core partof the HBP effort. An adapted strategy for data curation is needed, as the established workflows are not yetconsidering the integration of electrophysiological data.Another goal of the HBP is to provide a platform to facilitate collaborative research. For this the Collaboratory[2] has been set up - a web portal which provides a common online workspace (Collab) for all members of acollaboration team. The Collab combines tools which are developed in the context of the HBP platforms and bythis provides access to high performance computing (HPC), simulation tools and shared datasets. In particular, itis thought to act as a platform for interactive data analysis. Next to specialized tools which can be integrated ordeveloped for the Collab, generic analysis can be performed by using Python Jupyter Notebooks [3].MotivationThus, data used in the HBP must be prepared in two respects: (i) integration into the HBP databases and (ii) usein analysis processes on the Collab. Due to the diversity of data (types) in electrophysiological experiments,standardized data and metadata models, and tools operating on these models, have only started to be developed[4,5]. A crucial step in further advancing and disseminating these efforts, and to ensure that individualcomponents can be efficiently linked, is to embed these tools into workflows that recreate the actual scientificroutine of a research project.MethodsHere we consider the combination of 4 open source projects attempt to address these issues:• Neo provides a generic standardized representation for electrophysiological data, which is able tointerface with a range of electrophysiological data formats [6].• The Electrophysiology Analysis Toolkit (Elephant) offers methods ranging from the analysis of spikedata to popluation signals, e.g., local field potentials. Elephant is based on the Neo data representationformat [7].• The open metadata Markup Language (odML) is based on XML and offers a hierarchical structure tostore metadata related to electrophysiological experiments [8].• NIX is a file format designed to combine electrophysiological data and metadata in a single,standardized format [9], and is linked to both the Neo and odML data models.Results & DiscussionHere we show in 3 stages how these open source programs can interact to form a structured, comprehensibleworkflow for electrophysiological spike data. Firstly, we demonstrate the loading of data and metadata and theirintegration into a single data representation. For this we start with the conversion of the raw data into a Neoobject, which is then further annotated with metadata information. To obtain the metadata information, primarymetadata are first converted to the odML format using the odMLtables tool [10], which is then queried for theannotating the Neo object. In a second stage, the final Neo object is saved as NIX file, which preserves the data-metadata relations formed in the Neo structure. In a last stage, the data structure from the NIX file is used forexemplary analysis of the spiking activity using Elephant.In additon to the implementation of such a workflow in Python for use on a local machine, we demonstrate thesetup of the workflow on the Collaboratory of the HBP and indicate how the interaction of multiple collaborationpartners can benefit from the workflow realized in this setting. In addition, we discuss how the data, in particularthe odML-based metadata, can be used for integration in the registration processes developed by theNeuroinformatics platforms.References:[1] https://www.humanbrainproject.eu[2] http://www.collab.humanbrainproject.eu[3] http://jupyter.org/[4] Denker M., Grün S. (2016) Designing Workflows for the Reproducible Analysis of Electrophysiological Data. In: Amunts K., Grandinetti L., Lippert T., Petkov N. (eds) Brain-Inspired Computing. BrainComp 2015. Lecture Notes in Computer Science, vol 10087. Springer, Cham[5] Zehl, L., Jaillet, F., Stoewer, A., Grewe, J., Sobolev, A., Wachtler, T., Brochier, T., Riehle, A., Denker, M., & Grün, S. Handling Metadata in a Neurophysiology Laboratory. Frontiers in Neuroinformatics. 10, 26.[6] Garcia S., Guarino D., Jaillet F., Jennings T.R., Pröpper R., Rautenberg P.L., Rodgers C., Sobolev A.,Wachtler T., Yger P. and Davison A.P. (2014) Neo: an object model for handling electrophysiology data in multiple formats. Frontiers in Neuroinformatics 8:10: doi:10.3389/fninf.2014.00010[7] https://github.com/INM-6/elephant[8] Grewe, J., Wachtler, T., & Benda, J. (2011). A Bottom-up Approach to Data Annotation in Neurophysiology. Frontiers in Neuroinformatics, 5, 16.[9] https://github.com/G-Node/nix[10] https://github.com/INM-6/python-odmltables
|
The record appears in these collections: |