![]() ![]() For example, analysis results can motivate further collection of experimental data, whereupon it is clearly of advantage that they are made available once they are produced. In this case, scientific workflows allow to optimize and then more efficiently execute scientific processes (Ludäscher et al., 2009). Tasks of such a project workflow can interdepend: a further step of the local work depends on another operation that is remotely carried out. One challenging aspect of project workflows might concern immediate sharing of highly structured and voluminous data across labs. Projects with such multi-university collaborations benefit from well organized coordination of the participating specialists (Cummings and Kiesler, 2007). It tackles questions spanning disparate levels of organization such as genes, neurons, circuits, and behavior with a variety of methods including sequencing, electrophysiology, and computer simulations (Shepherd et al., 1998). In addition, the need for multi-university collaboration is particularly acute in neuroscience being a multilevel discipline. Ideally, peer-reviewed data should also be available for replication and re-analysis to test new hypotheses as knowledge progresses. Both measured and simulated data need to be stored in raw form, preprocessed, contextualized with metadata, organized to facilitate queries, and then analyzed to produce scientific statements. ![]() Science today deals with a “data deluge” caused by the widespread use of high-throughput sensors in experiments, and the ever more complex simulations afforded by increased computational power (Moore, 1965). We are motivating our solution by solving the practical problems of the GinJang project, a collaboration of three universities across eight time zones with a complex workflow encompassing data from electrophysiological recordings, imaging, morphological reconstructions, and simulations. ![]() We provide an implementation that relies on existing synchronization services and is usable from all devices via a reactive web interface. Since data sharing is cloud based, our approach opens up the possibility of using new software developments and hardware scalabitliy which are associated with elastic cloud computing. Using NeuronDepot is simple: one-time data assignment from the originator and cloud based syncing-thus making experimental and modeling data available across the collaboration with minimum overhead. The main drivers for our approach are to facilitate collaborations with a transparent, automated data flow that shields scientists from having to learn new tools or data structuring paradigms. We report here (1) a novel approach to data sharing between collaborating scientists that brings together file system tools and cloud technologies, (2) a service implementing this approach, called NeuronDepot, and (3) an example application of the service to a complex use case in the neurosciences. Neuroscience today deals with a “data deluge” derived from the availability of high-throughput sensors of brain structure and brain activity, and increased computational resources for detailed simulations with complex output. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |