A unique collaborative project uses a variety of tools and technologies to connect the neuroscience community and promotes a “wisdom of the crowds” approach to solving challenging problems in brain research.
In November 2009, a team of researchers from the University of California at San Diego created the Whole Brain Catalog (http://hcp.lv/hyqaON), a ground-breaking, 3-D virtual environment aimed at connecting “members of the international neuroscience community to facilitate solutions for today’s intractable challenges in brain research through cooperation and crowd sourcing.”
Co-creators of the catalog, Mark Ellisman, PhD, Stephen Larson, and Maryann Martone, PhD, of the National Center for Microscopy and Imaging Research at UCSD, discuss the massive, open source, open-access database of brain imagery.
What is the Whole Brain Catalog?
It is one piece of an assembly of systems and software meant to make it easy for anyone interested in the brain to find their way into a treasure trove of information and to help assemble all information about the brain in a way that makes it assessable and freely available. The Whole Brain Catalog itself should be thought of as a 3-D window into a community-assembled asset pool of information. The information resides in databases that are seen through the catalog and connected by the Neuroscience Information Framework (http://hcp.lv/iivLtw
), which is like Wikipedia on steroids. With regards to how effectively the information on the brain is linked, you have data and search capabilities that bring all kinds of information that’s accruing from researchers around the world. Because most of the imagery that’s obtained is static, the system allows one to bring in animations that are the results of simulations using everything up to super computers to generate them.
The Whole Brain Catalog is the flagship product of the Whole Brain Project (http://hcp.lv/gQiEgp
). We like to think of it as Google Earth for the brain. There’s a whole community, computational neuroscience, which works on the problem of how to make predictive models of systems of neurons. Many of those models are generally developed with some relationship to the actual cells, but not all is done in a full anatomical context, like where exactly those cells should be in relation to the rest of the brain. With the Whole Brain Catalog, you can take either pre-assembled simulations or create new simulations and see, for example, neuron patterns firing.
It’s an idea that we’ve had around for some time—to create a way of representing the brain that capitalized on special maps using spatial navigation technologies that are frequently used in maps of the Earth or cities, and then to link that information over the open Internet. We previously worked on this for a project called the Smart Atlas (http://hcp.lv/fHFHhj
). The Whole Brain Catalog took that to a new level by basing the front end on an open-source game engine in an environment where you could recruit others to build components so that the work to build and populate it could be done openly through crowd-sourcing. In our field, there’s not a culture of sharing research data. We’re trying to use this project to make available as much data as possible and to encourage scientists to convert from holding data closely within their laboratory to making data available for reanalysis openly.
The Whole Brain Catalog is meant to be a platform for sharing fairly complicated types of data, unlike a wiki where somebody comes and writes some text. We do maintain the NeuroLex (http://hcp.lv/encgfB
), which is a wiki that includes all of the neuroscience concepts that are needed to describe this data. Scientists can use it to relay and describe data to each other. Unlike a traditional wiki, it’s more structured and amenable to computational processing. Plus, the Whole Brain Catalog is dealing with special graphics and 3-D representations and many, many different data types. It’s not as easy as going to a wiki page and opening a text editor.
Who contributes the images?
Basically, the entire neuroscience and computational neuroscience communities are potential contributors. So far, there are several thousand datasets in, so I can’t tell you where they all come from.
One example would be the core funding that we’ve received from the Waitt Foundation (http://hcp.lv/eFgp6D
). Ted Waitt, co-founder of Gateway Computers, saw what we were interested in doing, thought it would be a game-changer, and gave us the assets to build the first instance of this. Also, the International Retinal Research Foundation (http://hcp.lv/dLxwKB
) has given us money to add the eye and the optic nerve, with the idea that we would reach out to researchers working on glaucoma or other diseases that affect the eye and bring those data in. We’ve already successfully brought in—with Johns Hopkins University investigator Nicholas Marsh-Armstrong—datasets that will contribute significantly to our understanding of the degenerative process associated with glaucoma. Other researchers interested in glaucoma will be expected by the foundation to start contributing their data to this environment.
There are similar contributions for things that are not disease-related and extremely basic. For example, our work with the Gage Lab (http://hcp.lv/hgHngA
) looks at how the nervous system, in places where new cells are born in the adult, take those cells and allow them to wire up so that they contribute to some new functional capability.
There are researchers working on different key areas of the brain and the cortex, for example, trying to understand how the visual information is represented in patterns of wiring within the cerebral cortex. A group at Harvard is right now contributing probably the largest single dataset taken on this very large area and at very high resolution.
There are very few places where one can explore data across so many scales, because it’s very expensive to store date, especially when you want to access large data, but we’re trying to do so and provide efficient ways of letting those data sets be viewed on a thin client.
How many images and videos are contained within the catalog?
Around 3,700, but when you load it on your system, or when you open the Web version that is just launching, you actually don’t have all that data loaded into your system; you pick the pieces. Since the system looks across many data sources, as long as people link their information in from other repositories, the number can grow astoundingly, rapidly.
The Neuroscience Information Framework (NIF) project that Mark mentioned has really been charged with accounting for how many of these electronic research portals are scattered around the globe that would be relevant for neuroscience. The NIF currently provides access to 60, with about 30 million pieces of data, not all directly related to the brain, but to genetic pathways that are certainly active there. And the number of databases that are potentially there to link to the Whole Brain Catalog is well over 1,000.
The Whole Bain Catalog should cause, once people explore it more effectively, many of the sorts of datasets that the NIF project has made visible to be spatially integrated. Right now there hasn’t been sufficient work done to curate them into a small street corner of the brain. The tools of the catalog will allow you to grab the dataset and then morph it, warp it, and fit it. And then it becomes available to somebody else as something that’s been fitted in like a puzzle piece.
What’s the user experience like?
We have areas of the catalog that are marked, essentially like bookmarks on your browser, where we house a significant amount of data. You click on one of the bookmarks, and it takes you to a spot in the catalog. There’s also a browser that’s like the layers of Google Earth; as you launch and see the different datasets, you can see the different layers. There’s a layer for all the cells in the catalog, and there’s a way to see things that are smaller. When each of those items load up, you can double-click on them in the data browser, and it will take you directly to that spot in space where they’re located. Additionally, we have a search functionality that allows you to look for things that have been tagged inside the catalog. There’s currently a preview of our Web version that’s a completely browser-based tool. It previously required a download; we’ve worked hard to put that inside the browser.
How can practicing psychiatrists and neurologists benefit from the Whole Brain Catalog?
The Whole Brain Catalog started out with the rodent brain as it’s main center point, because rodents are heavily studied inside experimental neuroscience. Practicing psychiatrists and neurologists, I believe, are familiar with the anatomy of the rodent and do pay attention to the literature that exists inside studies that are done on animals. To the extent to which animal science is valuable to them in their understanding of the human brain and the way the human brain works, this a valuable resource for seeing what those structures actually look like when they’re inside the brain. We do aspire to include the human brain as part of the catalog, and I think that’s going to be a really exciting opportunity that crosses over even better with psychiatrists and neurologists.
To the extent that you’d think of the Whole Brain Catalog as a portal to vast amounts of neuroscience data, NIF covers all of neuroscience, and so it offers access to deep databases across the literature and to different registries. Coming from the Whole Brain Catalog, one can very easily end up inside the NIF offerings.
It’s much more focused and direct access to more relevant information than a patient or a practicing psychiatrist would obtain just going to Google or Bing, because the NIF has very neuroscience-smart ways of organizing what you see when you instigate some sort of a complex query.
We also try to go after the content that’s not really well indexed by search engines. In the NeuroLex wiki, we expose all of the concepts that neuroscientists use for search and use those inside the NIF search engine to help us grab information a little bit more effectively from these resources than otherwise is possible.
What have you been able to learn because of the existence of the Whole Brain Catalog?
Recently, we collaborated with a laboratory that’s doing basic science work on the sense of smell. When you take a whiff of a nice glass of wine, there are probably hundreds of molecules that act like keys connected to locks inside your nose that are little receptors that form a map in the first level from your sense of smell into the brain. That means that the same type of molecule will go to a specific cell in your brain, and that forms a map that tiles the space of sense of smells. What this research was interested in is whether that map goes one more level into the brain.
You can think of the brain in terms of sensory systems as different layers of input. The first layer is a map; now what does the second level look like? They used some very sophisticated imaging techniques with which they can make a single cell in the brain turn into a fluorescent green and look at it under the microscope and reconstruct the pathway of that cell.
They’ve used the Whole Brain Catalog to take the results of these imaging experiments, put them into 3 dimensions, and then manipulate them so they can essentially collect, across individual animals, a single picture of what these cells look like in 3 dimensional space. To do that, they have to take 3D neurons and put them into the same space. At that point, you can start to do computational analysis. They looked at different endpoints from these neurons to begin to understand overlap. They found that it’s a lot less obvious of a map at the next level. In fact, they don’t find the same type of regularity that we see below, but that’s important to know because it means that there’s some other function at work there.
Another example is the project we undertook because the International Retinal Research Foundation needed the addition of the retina and optic nerve to the Whole Brain Catalog, which lead us to understand a previously unknown mechanism by which non-neuronal cells in the optic nerve, astrocites, appear to take up the debris that accumulates in neurons during their life. Instead of those materials, which in the conventional view of how cells work would be handled by machinery in the cell body by the nucleus to digest and clear old proteins, it looks like these optic nerve cells have an accessory mechanism just behind the eye to clear that material, and that was not discovered until we took it upon ourselves to put data related to these experiments, and new kinds of data, and put them in the catalog. It shows the power of cooperative work, collaboration, and data sharing.