Sat. Oct 1st, 2022
The most complete brain map ever is here: it

When asked what’s so special about it Drosophila melanogaster, or the common fruit fly, Gerry Rubin quickly lands a part. Rubin has poked and prodded flies for decades, including leading the effort to sequence their genome. So allow him to count their earnings. For example, they are expert navigators, who fly around without crashing into walls. They also have great memories, he adds. Deprived of their senses, they can find their way around a room – just as you, if you were suddenly blindfolded, could probably escape through the door you last entered.

“Fruit flies are very useful,” he estimates. And all that ability, while present in a poppy-seed-sized brain, involves some neural circuitry similar to our own, a product of our distant common ancestor. That’s why, as director of the Janelia Research Campus, part of the Howard Hughes Medical Institute, he has spent the past 12 years leading a team mapping the physical wiring of the fly brain down to the very last neuron.

Janelia researchers on Wednesday announced an important step in that search by releasing a wiring diagram of the fly’s brain that contains 25,000 neurons and the 20 million connections between them. The so-called ‘connectome’ corresponds to the fly’s hemibrain, an area about 250 micrometers across – the size of a dust mite or the thickness of two strands of hair. It is about a third of the total fly brain and contains many of the critical regions responsible for memory, navigation and learning.

Rubin hopes wiring diagrams like this one showing neurons involved in navigation will give researchers a better idea of ​​how brain circuits work.
enlarge Rubin hopes wiring diagrams like this one showing neurons involved in navigation will give researchers a better idea of ​​how brain circuits work.

Flyem | Janelia Research Campus

Researchers like Rubin believe that a physical blueprint of the brain could become a fundamental resource for neuroscientists — to brain science what genome sequences have done to genetics. The argument is that to get anywhere in understanding brain circuits, you first need to know what the circuits are and what kind of cells they put together. That physical schema becomes a roadmap for all kinds of research, Rubin says, from understanding the role of the brain’s wiring in psychiatric disorders to how our brains store memories.

Of course, it would be nice to explore those questions with an all-human connectome. But that is still a long way off. Completely analyzing even the smallest amount of brain matter costs an enormous amount of time and money.

Hence the brain of the humble fruit fly, with a millionth the number of neurons of ours. Drosophila is only the second adult animal whose brain circuitry has been mapped at this level of detail, after the nematode C. elegans in 1986. That task was much more modest. The entire nervous system comprised 302 neurons and 7,000 connections — small enough for researchers, with enough effort, to get the job done by physically shearing layers of cells, printing images taken with an electron microscope, and trace with colored pencils. The complexity of the fly brain is two orders of magnitude greater, so the gap of three decades to get it done.

“It’s a milestone,” said Clay Reid, a neuroscientist at the Allen Institute in Seattle who has worked on creating a similar map for a cubic millimeter of mouse brains. To the small community of researchers who have spent decades building connectomes, the emergence of these first large-scale data sets feels justified, he says. “At first people thought we were certifiable. And if we weren’t crazy, we’d be boring.”

Reid and Rubin’s fellow neuroscientists wondered whether such schemes would be useful, given the sheer number of unknowns about how neurons work. You might end up with the physical structures, but little understanding of the neural activity going on there. The rest thought the whole undertaking was unfeasible. In 2004, researchers at the Max Planck Institute in Germany had demonstrated automated methods that could analyze images of neurons produced by electron microscopes — a process known as segmentation. It was a huge improvement over tracing neurons by hand. But even then, completing the entire flybrain connectome would have taken 250 people working for two decades, Rubin estimates.

Google's Algorithms "paint" black-and-white images of neurons to give a clearer picture of where the cells begin and end — a process called segmentation.
enlarge Google’s algorithms “paint” black-and-white images of neurons to give a clearer picture of where the cells begin and end — a process called segmentation.

Flyem | Jamelia Research Campus

Rubin was undeterred and bet the technology could be accelerated. The team initially focused on improving data collection methods using electronic microscopy. To get the full neuron-by-neuron map they were hoping for, the researchers had to develop new computational techniques to produce clearer, denser three-dimensional images. The process involved cutting the brain into 20-nanometer plates, then imaging them continuously for months in an undisturbed environment. A small error in one part of the imaging can cause ripple effects across the entire dataset of the connectome.

But the real bottleneck remained in the process of understanding those images: the segmentation problem. Viren Jain, a former Janelia lab manager, worked on that same problem at Google, using a machine learning technique called flood-filling networks. While previous methods consisted of detecting boundaries between neurons and then grouping related pixels together, the new method combined those steps to fill in neurons one at a time — “like making a painting of the image,” says Jain.

To train its machine learning algorithms, Google needed data — images of neurons filled in by humans — that Janelia could provide. It also needed a human fact check. Back at Janelia, after the computers filled in the neuron images, a team of about 50 proofreaders looked at the algorithm’s results, looking for erroneous shapes and connections. “Computers can’t do all the work,” adds Jain.

Now that the data is publicly available, it remains to be seen how researchers will use the scheme. While researchers could previously zoom in on interesting brain circuits and map those neurons, this has been costly, says Reid, a researcher at the Allen Institute. He hopes a complete map will help researchers see distant connections that would otherwise be overlooked. It is also potentially more efficient. “Before, every question required a difficult experiment. But now it’s a computer question,” he says. “There is no comparison.”

That’s an exciting prospect for researchers like Karla Kaun, a professor at Brown University who studies the effect of drugs and alcohol on memory formation. Having a detailed map of the hemibrain is important, she says, for understanding the nuances between the circuits involved in the extremely long-lasting, intense memories and more typically long-term memory. She would like to see that data combined with other methods that can be used to cheaply compare structures in brains within the same species. That could potentially show how differences in physical structure contribute to disease and behavior.

A few other connectome projects are grinding ahead. Google is working with researchers from Max Planck to analyze circuits in the brains of songbirds involved in song learning, and with Harvard to study a tiny human monster. “It’s the opposite of the fly project, just a millionth of a whole human brain,” says Jain. Still, it may be a petabyte of data. Later this year, Reid expects to release an even larger dataset that matches the cubic millimeters of the mouse brain his team has imaged, as part of a project funded by IARPA.

One advantage that independently funded places like Allen Institute and HHMI share is that they can make these long-term bets. “I was kind of a venture capitalist here, letting the money flow for 12 years and making sure no one killed anyone,” Rubin says. Janelia spent $40 million on the project, not counting Google’s contribution, for which the cloud computing budget alone would run into the millions. Janelia has an ongoing budget of $5 million a year to map the entire nervous system of both a male and female fly.

Rubin hopes that effort will pay off in time. “I’ve been through the genome projects,” he says. “I can remember in the 1980s when people said you just get a series of AGCTs and you don’t know how to interpret the data.” We still don’t know how to read that series, not even close, but we’re making progress. And the cost of sequencing genomes has dropped significantly in the process. “Anyone who thought the genome project was stupid is now admitting it was worth every penny,” he adds.

Still, it’s unclear who will pick up the tab on future connectomes, especially with the most interesting brains that are orders of magnitude larger than the fly’s. Rubin is aiming for a full-fledged attempt to map the mouse’s 75 million neurons. The cost might be $500 million, he muses, assuming the tools accelerate maybe two or three orders of magnitude. But that was the case when he embarked on his own brain mapping journey. “We’ve shown people that it’s achievable,” he says.

This story originally appeared on wired.com.

By akfire1

Leave a Reply

Your email address will not be published.