A Neuroelectromagnetic Data, Tools, and Compute Resource
This NEMAR portal for human electrophysiological data supports sharing and in-depth analysis of de-identified human neuroelectromagnetic (NEM) brain data – scalp EEG, its magnetic counterpart, MEG, and, intracranial iEEG and ECoG.
Most neuroimaging data archives function as data libraries allowing users to download one dataset or study at a time. Typically, these serve users prepared to re-analyze data from a study of interest using their own tools and computing environment. The current era of machine learning and "big data" analysis calls for the creation of integrated DAta, Tool, and COmpute Resources (DATCORs) that integrate archives of sufficiently annotated data, stored in a standard format, together with a flexible data search facility and readily extensible data analysis tools, all within or directly connected to a powerful and widely available computing environment.
The NEMAR archive will allow users to select and then analyze data directly using the adjacent Neuroscience Gateway (NSG, www.nsgportal.org) platform, which offers a variety of neuroscience tool environments (including MATLAB/Octave, Python, R, Freesurfer, EEGLAB, PyNN, Tensorflow) for processing data on the XSEDE (Extreme Science and Engineering Discovery Environment) high-performance computing network (XSEDE.org). Data uploaded to NEMAR and then or later made public by its authors will also reside on the OpenNeuro archive (www.OpenNeuro.org). NEMAR data will be stored in the BIDS formats (Gorgolewski 2016). NEMAR will encourage the use of the Hierarchical Event Descriptor (HED) standard (Bigdely-Shamlo 2016) within BIDS to more precisely describe experiment events. For this purpose, tools to archive and search NEM data using the forthcoming HED v3 will be released and integrated into NSG tool environments including EEGLAB.