Allen Institute for Brain Science expands its atlas collection.
By Jim Kling
Oct. 8, 2008 | Founded in 2003 with $100 million in seed money from Paul Allen, the Seattle-based Allen Institute for Brain Science (AIBS) is a short walk from the center of the well-known art district of Fremont. Institute offices overlook a portion of the canal that links Lake Washington and Lake Union, treating employees to the occasional views of kayaks and other watercraft.
There’s not much time for people watching at AIBS, though. The institute created its flagship Allen Brain Atlas (ABA) to fuel neuroscience discovery. Since its initial publication in September 2006, the ABA has been a boon to neuroscientists worldwide. The project was ambitious: it generated a 3-D map of gene expression at cellular resolution in the mouse brain that can be visualized by specific genes and through anatomical reference points. The atlas is freely available online and has been cited more than 150 times, says Kelly Overly, research alliance manager for AIBS.
Last July, AIBS published the first results of a follow-up project—the mouse-based Allen Spinal Cord Atlas. Currently, about 2,000 genes have been mapped, with the institute projecting genome-wide coverage by the end of the year. It too can be searched and sorted by gene, age, expression, and anatomical structure.
Two other projects are farther from completion. The ABA Human Brain will be complete in about four years. The ABA Developing Mouse Brain will follow gene activity across different stages of development between birth and adulthood and will be completed in two years. These projects represent a major expansion for AIBS. To drive them, Overly anticipates a 50% increase in the scientific workforce by the end of 2008.
Cutting Edge Informatics
AIBS’s researchers largely use off-the-shelf technology—microscopes, cameras, slide carriers, automated stages—but focus on developing custom software to handle parallel imaging systems. The ABA employed ten systems in parallel running 24/7 with little operator control.
“It was largely a matter of putting things together in a way that hadn’t been done before. The challenge, like with any biological system, was to develop algorithms that can be robust but still account for the inherent variability in biology,” says Overly. The researchers employed a bright field microscope for greater resolution with 10x lenses. Each image section was broken into one hundred tiles that yielded cellular-level resolution.
The greatest hurdle was the downstream processing, Overly says. The team wanted to encode, align, and package the data in a way that would allow users to visualize it at will, like a virtual microscope. They also wanted users to have good flexibility in mining the data.
The spinal cord project required some modifications to the LIMS system, as well as new methods for sectioning and slide preparation. Moreover, the imaging software had to be taught to recognize the new configurations. But the basic technology is the same. “We’re just tweaking the system to account for different-sized tissue specimens and different signal characteristics due to the nature of the tissue,” says Overly.
The net result is a publicly-available web site that includes some powerful tools. The Brain Explorer desktop utility allows users to explore 2-D data sets in a 3-D representation. Researchers can also do searches by anatomical structure. Says Overly: “One of the most powerful tools is the search function ‘genes like me.’ Say a user is interested in the pattern of expression [of a particular] gene in the hippocampus. You can seed the system with that gene and tell it to show all other genes with similar expression patterns in the hippocampus. That allows the user to pull out groups of genes that might be [related] to one another by virtue of proximity. Having the whole database of genes and being able to combine spatial information with the gene data gives them a jump start in their research.”
Another tool, the Anatomic Gene Expression Atlas, gives an unbiased view of the brain that is driven by gene expression, according to Overly. “It allows you to ignore every classical marker of neuroanatomy—you’re just looking at gene expression and seeing which areas are most similar to each other. It shows a (genetic) organization of the brain. What you find is that in many cases the gene expression patterns line up reasonably well with the anatomical divisions that have been established using other methodologies during the years. If you go to higher and higher resolution, gene expression might be able to inform some of the more fine boundaries within the brain. It allows users to understand the organization (of the brain) in a different way.”
The anatomical structure of the spinal cord is simpler than that of the brain, so the spinal cord atlas doesn’t include a 3-D reference atlas. “It’s a bit more streamlined database with images of gene expression,” Overly says. ABI plans to release anatomical reference images that can be used with the data.
The two atlases are extremely valuable to neuroscientists, says Jane Roskams, associate professor of neuroscience at the University of British Columbia: “The strength of a tool like that isn’t just that it helps you find clues to pathways, but it also tells you which pathways are not important (to your system). It’s easy to overlook that because it doesn’t end up in a published paper, but it gets you ten steps further (in your research) than you would have been if you didn’t have access to that information.”
Roskams helped AIBS locate spinal cord experts to advise on the project. She soon realized how little is known about the system. “We don’t know the function of many of the cells,” she says. “We know the region they’re in—regulating pain, or sensory signals, or near motor neurons that drive large cells—but we’ve only scratched the surface of understanding how they work together.”
The developing brain atlas will complement the other gene atlases. Roskams expects it to provide a road map to guide spinal cord regeneration researchers, for example, while the brain and spinal cord atlases can act as a measuring stick for success. “On the one hand [the developing brain atlas reveals] the programs you need to follow to reconnect areas of the spinal cord [following injury], and on [the spinal cord atlas] is a readout of what you should expect. It helps us to know whether or not we’ve reached the end zone,” says Roskams.
The projects aren’t the only ones of their kind, but AIBS’ resource base and institutional focus puts it in a unique position, Overly says. Other projects do not have the breadth and scope of the AIBS projects. “It’s challenging because everyone has different standards, and the volume of data produced is tremendous,” he says.
Roskams appreciates the effort. “It’s a long and tedious road to get that data and do it well, even for a single set of genes. They’re taking the tedium out and giving [the data] back” to the research community, she says. Still, Overly hopes that the research community will play an active role. The projects are designed to be open access. Users can also access the data itself, including the 3-D coordinate system and XML files of gene expression data. “We’re working hard to make as much of it open as we can,” says Overly. “Users can bring the data in house in a way that they can create mash-ups or mine it and manipulate it, or develop their own technologies on top of it.”
This article appeared in Bio-IT World Magazine.
Subscriptions are free for qualifying individuals. Apply Today.