OpenScope is open for business.
The Seattle-based Allen Institute for Brain Science has taken a page from the playbook for the Hubble Space Telescope to create its latest channel for open-access neuroscience.
Like Hubble’s handlers, the institute is taking requests from researchers for access to its experimental platform for observing neural activity in mice.
Leaders of the project went so far as to consult with leaders in the astronomy community, particularly at the W.M. Keck Observatory in Hawaii, to learn how they divvy up telescope time.
“What we’re doing here has never been done before,” Koch said. “Rather than peering out at the blackness of the sky, we are peering inward at the brain, and we are doing it in such a way that the data is reliable, statistically valid and openly accessible.”
Jérôme Lecoq, the institute’s senior manager of optical physiology, told GeekWire that the name of the project is “really a play on words.”
OpenScope refers not only to the open-source nature of the project and the telescope-like mode of operation, but also to the fact that it serves as a “microscope that’s open to the community,” he said.
“It’s an important transition for neuroscience. … This is offering a new way,” Lecoq said.
The concept means that neuroscientists don’t have to work through the mechanics of setting up the infrastructure for mouse experiments, just as astronomers don’t have to build their own telescope to make world-class, Hubble-level observations.
“You can literally make a proposal, and in a year you have the data,” Lecoq said.
The observatory’s current focus is the mouse visual system. To run experiments, the institute’s researchers show mice different images or movies. The mice are genetically engineered to have neurons that light up when they’re active. That makes it possible for researchers to record brain activity in different circuits through tiny windows in the animals’ skulls.
Readings collected from more than 65,000 different neurons already have been deposited in the Allen Brain Observatory’s publicly accessible database. Lecoq said experiments typically make use of five to 20 mice, but “we can do much more than this.”
“For instance, our latest publication, which just came out in BioRxiv, involved 200 mice,” he said. “This is the advantage of the OpenScope, because we built this infrastructure.”
The first three pilot experiments taking advantage of the OpenScope approach are already under way, with full funding from the Allen Institute:
One team, led by University of Colorado biophysicist Joel Zylberberg, is studying how learning in the visual system is coordinated between different regions of the brain, and how closely that process aligns with learning methods used in artificial neural networks.
OpenScope “will help to explore how rodents learn when faced with unexpected circumstances, hopefully helping us to answer questions about deep credit assignment in the brain,” said team member Yoshua Bengio, an expert in machine learning at the University of Montreal.
Other team members include Timothy Lillicrap and Blake Richards.
One of the first experiments conducted using the Allen Brain Observatory studied how mice reacted to movies showing a wide variety of subjects, including the opening scene of an Orson Welles film titled “Touch of Evil.” A follow-up experiment led by the Allen Institute’s Anton Arkhipov and the University of Wisconsin’s Giulio Tononi flips the script.
This time around, mice are shown pictures of critters they fear (snakes and cats) as well as critters they like (“Crickets … they eat them,” Lecoq said). Then the pictures are subtly morphed to pinpoint which visual features the mice rely on to identify what they’re looking at.
Other research team members include Yazan Billeh, William Marshall and Willam Mayner.
The path of experience
Allen Institute researcher Marina Garrett is leading a team that’s investigating how experience and expectation influence the way the brain represents information using specific pathways. Garrett and her colleagues – Nicholas Cain, Rylan Larsen and Hannah Choi – are monitoring brain activity in the mice during visually guided behavior, to see how those different pathways come into play.
“Hopefully we’ll have all three experiments done by the end of the year,” Lecoq said.
Lecoq said the institute will issue its next request for OpenScope proposals sometime next month. “At this point, we’re going toward a yearly cycle,” he said. “Our goal for next year is to keep the same number of projects, and a slightly larger number of experiments. The following year, we’ll see how it unfolds.”
Like the other scientific projects that Paul Allen has backed, OpenScope stresses open science.
“I think the cycle will be that you first collect the data set, and then the person who submitted the proposal basically has access in real time, and then together we release the entire set,” Lecoq said. “If you want your research to have an impact, I think you have to share the data.”