Data management routines for reproducible research using the G-Node Python Client library

Abstract

Structured, efficient, and secure storage of experimental data and associated meta-information constitutes one of the most pressing technical challenges in modern neuroscience, and does so particularly in electrophysiology. The German INCF Node aims to provide open-source solutions for this domain that support the scientific data management and analysis workflow, and thus facilitate future data access and reproducible research. G-Node provides a data management system, accessible through an application interface, that is based on a combination of standardized data representation and flexible data annotation to account for the variety of experimental paradigms in electrophysiology. The G-Node Python Library exposes these services to the Python environment, enabling researchers to organize and access their experimental data using their familiar tools while gaining the advantages that a centralized storage entails. The library provides powerful query features, including data slicing and selection by metadata, as well as fine-grained permission control for collaboration and data sharing. Here we demonstrate key actions in working with experimental neuroscience data, such as building a metadata structure, organizing recorded data in datasets, annotating data, or selecting data regions of interest, that can be automated to large degree using the library. Compliant with existing de-facto standards, the G-Node Python Library is compatible with many Python tools in the field of neurophysiology and thus enables seamless integration of data organization into the scientific data workflow.

Microelectronics, bioinformatics and neurocomputation for massive neuronal recordings in brain circuits with large scale multielectrode array probes

“…In fact, conventional approaches are generally not optimized in terms of read and write routines and typically involve many intermediate steps that require direct conversions or external libraries to access the data. Approaches to solve the bottleneck of the computational overload were proposed (Mahmud et al, 2012;Sobolev et al, 2014), but their use typically requires a deep knowledge in software programming. Alternatively, in the last years there has been an increasing effort to identify more appropriate data formats and structures to handle the analysis and to share data and results.…”

Why sharing matters for electrophysiological data analysis

“…The approach taken by Sobolev et al (2014) is to provide a consistent structure permitting efficient access to the electrophysiological datasets. Python is cross-platform, and multiprocessor implementations exist.…”

Preparing Laboratory and Real-World EEG Data for Large-Scale Analysis: A Containerized Approach

“…G-NODE (Sobolev et al, 2014) provides a full data management system along with supporting tools for access, data-sharing, and analysis. The G-NODE (Sobolev et al, 2014) odML data model (Grewe et al, 2011) provides support for organizing study metadata as key-value pairs.…”

“…The G-NODE (Sobolev et al, 2014) odML data model (Grewe et al, 2011) provides support for organizing study metadata as key-value pairs. The companion Neo data format (Garcia et al, 2014) provides a flexible method of manipulating arbitrary physiological data.…”