Value Added by Data Sharing: Long-Term Potentiation of Neuroscience Research

Neuroscience research is resulting in an enormous accumulation of complex data. The explosive growth of bioinformatics demonstrated that sharing and reanalysis of digital information can lead to exciting opportunities for scientific integration and discovery. In striking contrast, broad-spanning issues (data heterogeneity, privacy regulations, etc.) have so far limited the reanalysis of brain and neural data. Nevertheless, data sharing is possible in neuroscience as well, and the ever-increasing power of affordable hardware and software has contributed to recent spiking interest in this topic. The Society for Neuroscience (SfN) has recently established a Neuroinformatics Committee and launched the Neuroscience Database Gateway, which is already widely used by the neuroscience community (Gardner and Shepherd 2004). Funding agencies, such as the National Institutes of Health (NIH), have data sharing policies, several journals require data deposition at the time of publication, and a number of books, journals, and reviews have focused on this topic in the past few years (e.g., Eckersley and the OECD Working Group on Neuroinformatics 2003). For the 2007 SfN Annual Meeting, we have organized a Satellite Symposium entitled “The Rhyme and the Reason of Data Sharing” to present a series of recent success stories on data sharing and reuse in neuroscience. With examples covering a vast array of techniques, topics, and scales, the symposium was designed to appeal to both neuroscientists who could share their data and to those who could reuse and reanalyze shared data. The scientific session consists of six short presentations ranging from molecular approaches to whole-brain imaging (described in more detailed below, in the order of presentation) followed by a panel discussion. The symposium is sponsored by four NIH Institutes (NINDS, NIMH, NIDA, and NCRR) and we have secured participation commitment of several prominent neuroscience leaders, including David van Essen, SfN President, and Tom Insel, NIMH Director (delivering the opening remarks), Nora Volkow, NIDA Director (moderating the panel discussion), and Story Landis, NINDS Director (providing the closing remarks). Starting with “Multiple Reuses of in vivo Multi-Unit Recording Data”, the focus of Ken Harris (Rutgers University, Newark, NJ) is on electrophysiological time series (Harris 2005). Modern neurophysiological technology allows for simultaneous recording of hundreds of neurons, offering an unprecedented opportunity to study the function of neural circuits. The richness of this data (gigabytes per day in each lab) has fundamentally changed the work of physiologists. While a few decades ago the bulk of time was spent at the bench, now a single experiment is typically followed by months of computational analysis. Moreover, one data set can be suitable to address multiple scientific questions, which are often quite different from those that prompted the original experiment. In a recent example, Steuber et al. (2007) reanalyzed in vivo recordings in awake animals of other investigators (Goossens et al. 2004) to provide supporting evidence that the mechanisms they described using modeling and slice experiments also had behavioral consequences. More generally, strategies encouraging theorists to mine existing data Neuroinform (2007) 5:143–145 DOI 10.1007/s12021-007-0009-0