rain maps seem to come out in rapid succession these days. They take various forms: a map for word concepts, a map of individual cells’ activity, a map based on the organ’s physical contours.
What they share in common is the aspiration to take the lumpy mass of the brain and categorize it, somehow, into useable areas — not unlike the textbook brain images with their colored denotations of “occipital lobe” and “frontal cortex.”
But these maps often come along with a problem: They may not sync up with the other maps. Now a group of scientists have managed to sync up two of the most commonly used types of brain maps — for gene expression and brain structure — and they’re releasing their methods to any and all in the scientific community.
First, the genetic maps. The nonprofit Allen Institute, based in Seattle, is a relative old-timer in the field of brain mapping, having published its first human brain maps in 2010. That map consists of six different human brains that were imaged using MRI then chopped up into 1-millimeter cubes. Those cubes were tested with probes for all of our 20,000 genes to measure the relative levels of those genes throughout the brains.
The end result, the Allen Human Brain Atlas, has been likened to a Google Maps for the brain — a resource scientists can use to determine where they’re at in the brain and what genes are important there.
When the creators of this atlas unveiled it, they weren’t totally sure how any of their maps would be used. “What specifically are people getting out of some of the resources that we’ve spent the last more than 10 years generating?” asked Amy Bernard, the project architect at the Allen Institute.
While Bernard has anecdotal examples that the maps are useful as encyclopedias for scientists browsing to see what’s out there — “the nugget you need to find to support what it is you’re asking,” as she put it — the institute does not systematically collect information to find out about more interesting uses.
But in July, a new study caught staffers’ attention. Researchers at the University of Cambridge had melded together the Allen Human Brain Atlas with MRI images of living peoples’ brains.
That seemingly simple task had big implications: The gene map could be extrapolated to you, me, or any other living subject, despite the quirks of how our individual brains are shaped. That could allow for whole new types of studies of how our brains’ gene expression correlates with our brain anatomy, behavior, and disease risk.
The Cambridge data comprised images of brains from 300 young people aged 14-24. The researchers divided up these subjects into cohorts and compared between the age groups, something that can’t be done with the small sample in the Allen atlas.
And in this case, the maps weren’t used to answer a question, but rather to find the question in the first place. It’s what’s called an “agnostic” approach. “Rather than limiting yourself to saying, well, I need to have a really brilliant ‘a-ha’ moment to think of the perfect experiment, you let the data speak for itself,” Bernard said.
The Cambridge team — led by Kirstie Whitaker and Petra Vertes — looked at two things: the amount of white matter in teenagers’ brains (via MRI) and which genes are expressed in the areas where the white matter is changing the most (via the Allen atlas). They wanted to see what those genes could tell them about diseases like schizophrenia, which emerges during the teenage years.
“Most of the time what you’re measuring with MRI are large-scale things that you can’t really link to small-scale biology,” like gene expression, said Vertes. But by combining the two maps, they were able to do just that. And ultimately they demonstrated a direct correlation between changes in the structure of teenagers’ brains and genes that have been linked to schizophrenia.
For researchers studying teens, that could mean a better insight into the causes of schizophrenia. But for any researcher using MRI, it means that the Allen maps have a whole new potential.
Other labs could soon run similar experiments. The Cambridge team had to manipulate large datasets from the Allen Institute to get it lined up with their MRI data. In the process, they created a software package that could map any kind of MRI data to the Allen Institute’s map. The software is available for free online — meaning that researchers in any lab with an internet connection can draw upon the Allen data to scour for new genetic findings associated with their own back-catalogs of MRI data.
“There are great strengths of using MRI. It’s non-invasive, you can do it on real, living people, you can ask them about their experiences, you can ask them to do particular tasks,’” said Whitaker. Indeed, the team’s next project is to link fMRI, which measures brain activity rather than brain anatomy, to the Allen atlas.
Of course, there are limits to what you can find by taking this approach. The Allen map doesn’t “by any means cover the brain comprehensively,” said Matthew Glasser, a neuroscientist involved in a separate brain mapping project at Washington University in St. Louis.
And because every brain is different, it’s not always easy to know for sure that a region in one map is the same as a region in another. In Glasser’s lab’s recent work, they found that a region of the brain involved in language sometimes shifts to a different location in about 10 percent of people, sometimes even splitting in half with another brain region in between.
Vertes agreed, “There are so many problems with every type of neuroscience data. I think our only hope really is to cross-reference datasets at various scales and from various angles.”
“So this is what I think the future is: making lots of maps, but then relating those maps to one another.”