Integrated platform and API for electrophysiological data

Abstract

Recent advancements in technology and methodology have led to growing amounts of increasingly complex neuroscience data recorded from various species, modalities, and levels of study. The rapid data growth has made efficient data access and flexible, machine-readable data annotation a crucial requisite for neuroscientists. Clear and consistent annotation and organization of data is not only an important ingredient for reproducibility of results and re-use of data, but also essential for collaborative research and data sharing. In particular, efficient data management and interoperability requires a unified approach that integrates data and metadata and provides a common way of accessing this information. In this paper we describe GNData, a data management platform for neurophysiological data. GNData provides a storage system based on a data representation that is suitable to organize data and metadata from any electrophysiological experiment, with a functionality exposed via a common application programming interface (API). Data representation and API structure are compatible with existing approaches for data and metadata representation in neurophysiology. The API implementation is based on the Representational State Transfer (REST) pattern, which enables data access integration in software applications and facilitates the development of tools that communicate with the service. Client libraries that interact with the API provide direct data access from computing environments like Matlab or Python, enabling integration of data management into the scientist's experimental or analysis routines.

A scalable neuroinformatics data flow for electrophysiological signals using MapReduce

“…The primary features of the Cloudwave data flow include the use of Hadoop MapReduce and HDFS together with the flexibility to configure multiple parameters based on the availability of resources on a Hadoop cluster. This allows Cloudwave data flow to be deployed on different types of Hadoop clusters and to be used as a template to develop scalable neuroscience data processing data flow in many existing neuroinformatics projects, such as the GNDataPlatform (Sobolev et al, 2014a ).…”

“…The existing work on electrophysiological signal data management can be divided into two categories: (a) Data Representation Formats; and (b) Data Processing Tools. Although there is no existing standard for signal data representation, there are a large number of data formats developed by instrument vendors, researchers, and different neuroscience projects (Schlögl, 2010 ; Sobolev et al, 2014a ). Signal data representation formats need to meet the requirements of multiple stakeholders and address multiple challenges, including the inherent complexity of signal data such as different sampling rates and scaling factors (Schlögl, 2010 ).…”

“…The German Neuroinformatics Node (G-Node) integrates the NEO format with the open metadata Markup Language (odML; Grewe et al, 2011 ) to define the GNData format for use in a data management platform (Sobolev et al, 2014b ). In addition to the data format, the GNData signal data management platform aims to develop a common storage layer with a generic API based on Representation State Transfer (REST) web services for signal data annotation and access control (Sobolev et al, 2014a ). The GNData platform also uses the Hierarchical Data Format (HDF5) ( Hierarchical Data Format (HDF5), 2014 ), which has generated a lot of interest in the neuroinformatics community as a potential common representation standard, to store the signal data.…”

Framework for Collection of Electrophysiology Data

“…• INCF Dataspace (INC Working Group, 2013) enables interested research groups to connect to a distributed data file system based on iRods 2 . • G-Node 3 provides tools for data access, data management and data sharing, including a data sharing platform (Sobolev et al, 2014) based on common data models for electrophysiological data and metadata (Garcia et al, 2014;Grewe et al, 2011). To structure metadata, G-Node has developed odML, an XML schema for the creation of complex metadata structure in computer-readable format (Grewe et al, 2011).…”

Going beyond the current neuroinformatics infrastructure

“…Beyond neuroimaging, Sobolev et al ( 2014 ) present a data management platform for neurophysiological data, and Mouček et al ( 2014 ), and Tripathy et al ( 2014 ) describe techniques and methodologies for collecting and managing electrophysiological data.…”