By Esther Landhuis
Wandering the convention center among 30,000-plus researchers, students and vendors at the Society for Neuroscience annual meeting in San Diego last November, I struggled to wrap my head around a feature I was writing for this week’s Nature, on managing big brain data. Mice, molecular biology and cell sorting reigned supreme in my former life as a bench scientist. Neurons, brain imaging, terabytes — not so much. So when it came time to find an entry into the vast universe of the brain, I latched onto something that seemed small and manageable: the fruit fly.
Ann-Shyn Chiang of National Tsing Hua University, Taiwan, told the SFN crowd his team has spent a decade imaging 60,000 neurons in the Drosophila brain. The pictures produced 3D maps detailed enough to show which neurons control precise behaviors, such as shaking the head side to side (see video). But here’s the part that blew my mind: They aren’t even halfway done (flies have 135,000 brain neurons), and mapping the human brain with similar methods would take 17 million years!
Head shake behavior elicited by a 593.5-nm laser. Credit Po-Yen Hsiao and Ann-Shyn Chiang.
However daunting, efforts to map the human brain are intensifying around the globe. As researchers warm up to data sharing on a worldwide scale, a growing number of resources are emerging to help them analyze and share datasets of increasing size and complexity. The Neuroimaging Informatics Tools and Resources Clearinghouse (NITRC) and platforms such as OpenfMRI provide places to share brain imaging data. Open Neuroimaging Laboratory helps researchers find and curate publicly available imaging data and analyze it collaboratively. Some repositories are disease-specific, such as the European Medical Information Framework for Alzheimer’s Disease (EMIF-AD) and the US-based Global Alzheimer’s Association Interactive Network (GAAIN) and Autism Brain Imaging Data Exchange (ABIDE). NITRC’s image repository includes normal and disease-specific brain imaging data for researchers who want to conduct studies across specific subsets. For instance, you could search NITRC for data from “left-handed males ages 40-73 who have schizophrenia, autism, ADHD or Parkinson’s,” says Nina Preuss, programme manager for the NITRC, in Washington DC.
It’s unclear if these sites can be integrated. Moreover, some datasets have become so large it no longer makes sense to centralize them. The hope is that researchers “can at least come together and use common standards,” says Russell Poldrack, a neuroscientist at Stanford University in California. That way it should be “relatively straightforward to federate across projects.”
That’s what inspired Poldrack and colleagues to create the Brain Imaging Data Structure (BIDS), a standard for organizing and describing MRI datasets. From the BIDS website researchers can download instructions on how to name files and arrange them into folders to make it easier for other researchers to use the data. BIDS also specifies how to structure “metadata” — information such as instrument settings, what participants were doing while their brain was scanned, and other such details needed by researchers who want to analyze the raw data. In addition, the BIDS team has created tools to determine if existing datasets are BIDS-compatible and, if so, convert them to BIDS format. Similarly, Neurodata Without Borders (NWB) aims to create a unified standard for cell-based neurophysiology data. Such standards are still relatively new—organization structures often vary between different researchers in the same lab—but some groups are starting to transition to common standards, Poldrack says.
As neuroscientists increasingly share data and synchronize research efforts, they are starting to resemble the exquisitely coordinated organ they study.
You can read the rest of Esther’s feature here.