Courtesy of http://Neurodatabase.org
Compared to gene and protein data, which involve relatively simple sequences, the diverse nature of data characterizing brain activity is, well, mind-boggling. Data can range from functional magnetic resonance images of brain structures to action potentials. According to the Society for Neuroscience, more than 70 neuroinformatics databases and tools are in development right now; but most of them gather imaging, rather than action potential, data. A select few, however, gather neurophysiological data about intercellular interactions; one of those is http://Neurodatabase.org.
"The overall goal of http://Neurodatabase.org is to ... have recordings from cells in networks [so that we can] try to understand something about how neurophysiology is correlated with behavior, learning, sensation, and action," says lead database developer Daniel Gardner, professor of physiology and neuroscience and director of the Laboratory of Neuroinformatics at the Weill Medical College of Cornell University.
Gardner recently described Neurodatabase. org in an issue of Nature Neuroscience, dedicated to the 10th anniversary of the National Institutes of Health's Human Brain Project, which funded the database.1 Gardner explains in an interview that, whereas a figure in a journal is fixed in size and content, "if you have access to the dataset, you can expand or contract the sweep to look at the whole data record bunched up, or you can look at it expanded."
For example, published images of spike-train data are represented as dots or lines that blur together when neurons are strongly excited and transmit a lot of data. Thus only a qualitative impression can be made of the information communicated by a neuron. In contrast, http://Neurodatabase.org provides the actual spike timing to millisecond accuracy.
EASE OF USE
A chief objective has been to make the database easy to use and functional across a range of computer platforms and datasets. The data are submitted to http://Neurodatabase.org through Java tools that allow one to format the data to the database's requirements. A virtual oscilloscope allows users to view the data on their computers and change the sweep or gain as needed.
"http://Neurodatabase.org is one of the few databases that provide time-sequence data relevant to electrophysiological investigations. Most other neural databases are image atlases and the like," says Edward G. Jones, director of the Center for Neuroscience at the University of California, Davis. "There are only two or three others of this type. However, all but one [Gordon Shepherd's SenseLab site at Yale University] are restricted to data from limited research conditions," he adds.
A key issue with any database is making the data conform to a set of standards. Gardner and his colleagues make use of an XML-based markup language called BrainML, a data-description standard that allows transmission of data from http://Neurodatabase.org to and from its Java tools, and collaborations with the neuroinformatics community, such as SenseLab http://senselab.med.yale.edu/senselab.
"From what I've seen, http://Neurodatabase.org has succeeded well in allowing many different kinds of data to be entered and in allowing searches to explore and extract those data based on a wide variety queries," says David Van Essen, head of the Department of Anatomy and Neurobiology at the Washington University School of Medicine, St. Louis. "For neurophysiologists who are interested in that kind of data, I think it is a very strong start," adds Van Essen, who is a member of the Society for Neurosciences' Brain Information Group.
SHARE AND SHARE ALIKE
The success of http://Neurodatabase.org and other databases, of course, will depend on the willingness of scientists to share their data. An NIH policy issued October 1, 2003, mandates that all grantees with direct costs of $500,000 or more per year share their research data.
According to Van Essen, a mixed spectrum of attitudes exists towards data sharing. "Some scientists are hesitant to share it beyond the traditional methods," he notes. But for many scientists, the big impediment has been a lack of databases that are both available and reasonably easy to work with. "You don't just want to dump in piles of unmanageable data. But now we have a number of databases that are becoming heavily populated with a rich spectrum of data," Van Essen says.
So far, Jones has seen only one or two examples of the successful use of databases to obtain novel, original findings. Still, he says, the field is getting accustomed to using databases and to the concept of data sharing. "Some older scientists are perhaps resistant, but they will get over it."
- Emma Hitt