Janelia maps every neuron in an adult fruit fly's central nervous system
Two decades after Janelia proposed the project, the adult fly connectome is finished. The 1,000-fold efficiency gain credited with making it possible is measured against the institute's own 2008 forecast.
The Scientist · Science desk
What happened
- Janelia and its collaborators finished a wiring diagram of an adult male fruit fly's entire central nervous system: more than 166,000 neurons and millions of synapses across the brain, both optic lobes and the ventral nerve cord.
- The project was announced in 2008 and met with widespread skepticism from the neuroscience community, which considered an insect-scale reconstruction too slow and too expensive to attempt.
- Until now the only nervous system mapped in full belonged to the nematode C. elegans, 302 neurons that took more than a decade of manual tracing to reconstruct.
- The intermediate result was the 2020 hemibrain, a reconstruction of 25,000 neurons covering half the fly brain, which served as the proof that the imaging and segmentation approach would scale.
Why it matters
- What's now possible capability The transferable asset is the microscopy-and-segmentation pipeline, which other groups can inherit without repeating the fly, so whole-nervous-system reconstruction becomes something a laboratory can schedule.
- What it limits constraint The vertebrate targets Janelia names are transparent fish, so mammalian-scale mapping stays an extrapolation from a demonstration on 166,000 neurons in an invertebrate.
- Who has to decide decision Groups that built circuit models on the half-brain data now face a choice about whether to re-anchor those results to the complete central nervous system, including the optic lobes and nerve cord they previously lacked.
Next story · Science · yesterday
Shanidar A's shoulder-blade hole fits a wooden spear, Cambridge researchers sayConfirmed12 publishers
The 1,000-fold figure needs a denominator. When the work began, Janelia's own calculation was that a fly connectome built with standard methods would require 500 specialists working for a decade [11]. That is 5,000 person-years of tracing [17]. Divide it by 1,000 and you get roughly five person-years [20]. The institute states the gain as a fold-reduction in the time and resources needed to generate whole-brain connectomes and does not report the hours the finished reconstruction consumed [8], so the comparison is with a 2008 projection. Nobody ever ran the 500-specialist version.
Gerry Rubin, Janelia's head of biology and its founding executive director, put the bet in those same terms [13]. "It was us having the leap of faith that we could assemble an interdisciplinary team who would develop ways to increase the efficiency of generating connectomes by more than 1,000-fold. That was our key contribution, and without that, we could still be waiting," he said [12].
Two bottlenecks had to move together. Harald Hess's group refined and scaled customised electron microscopy until it captured high-contrast images of individual neuronal membranes at nanoscale resolution [4], and computational teams working with Google Research built machine-learning models that segment dense neural tissue and read the resulting image volumes automatically [5]. The full central nervous system holds about 6.6 times the neuron count of the 2020 half-brain [18]. Against the older reference point the jump is larger: roughly 550 times the neuron count of the nematode [19].
What the dataset holds is which neurons contact which, in one animal, an adult male [6]. Nothing in it describes how wiring differs between individuals or between sexes. The evidence for what such a map buys sits downstream, in the hundreds of follow-up studies of sensory and cognitive processing that the half-brain reconstruction set off in laboratories worldwide [15].
Janelia's vertebrate step is transparent fish: larval zebrafish and adult Danionella, aimed at models of complex behaviour and of neurodegenerative disorders [16]. I would expect those reconstructions to arrive faster than the fly's did, because the microscopy and the segmentation models already exist [4][5]. Whether the same pipeline reaches a mammalian brain is a separate claim, and the fish are the first test of it.
What to watch
- Whether the zebrafish and Danionella reconstructions publish with reported person-hours, which would test the 1,000-fold claim against a measured run instead of a forecast.
- Whether groups outside the Janelia and Google Research collaboration run the segmentation models on their own electron microscopy volumes.
- Whether a second fly central nervous system is reconstructed, giving the first measure of how much wiring varies between individuals.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence60
- Adoption40
- Hype gap+25
- Incentives55
- Confidence62
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Scientists mapped every neuron and connection in the adult male fruit fly's central nervous system, a complete wiring diagram of more than 166,000 neurons and millions of synaptic connections spanning the brain, both optic lobes and the ventral nerve cord.
- [2]
The connectome was completed by investigators at HHMI's Janelia Research Campus in collaboration with teams at the MRC Laboratory of Molecular Biology, the University of Cambridge, and Google Research.
- [3]
By developing high-resolution microscopy and AI-powered computational algorithms in collaboration with Google Research, investigators reduced the time and resources required to generate whole-brain connectomes by more than 1,000-fold.
- [4]
Under senior group leader Harald Hess, the research group iteratively refined and scaled customized electron microscopy techniques to capture high-contrast images of individual neuronal membranes at nanoscale resolution.
- [5]
Computational teams partnered with Google Research to build machine-learning models capable of segmenting dense neural tissues and interpreting vast image datasets automatically.
- [6]
The completed connectome is of one adult male fruit fly (Drosophila melanogaster).
- [7]
The ventral nerve cord is the insect counterpart to the vertebrate spinal cord.
- [8]
The HHMI Janelia release expresses the improvement as a more-than-1,000-fold reduction in the time and resources needed to generate whole-brain connectomes and reports no labour-hour or cost figure for the completed reconstruction.
- [9]
When Janelia scientists first announced their ambition in 2008 to chart every neuron inside the fruit fly brain, the project was met with widespread skepticism across the neuroscience community.
- [10]
Before this work the only fully mapped nervous system was that of the nematode Caenorhabditis elegans, an organism with just 302 neurons, whose reconstruction required more than a decade of tedious manual labour.
- [11]
When the project commenced, researchers calculated that assembling a fly connectome using standard methods would require 500 specialists working for a decade.
- [12]
"It was us having the leap of faith that we could assemble an interdisciplinary team who would develop ways to increase the efficiency of generating connectomes by more than 1,000-fold. That was our key contribution, and without that, we could still be waiting," said Gerry Rubin.
- [13]
Gerry Rubin is head of Biology, a senior group leader, and the founding executive director at Janelia.
- [14]
The scaling effort achieved proof of concept in 2020 with the release of the hemibrain, a detailed reconstruction of 25,000 neurons across half of the fly brain.
- [15]
The hemibrain catalysed hundreds of follow-up studies into sensory and cognitive processing across global laboratories.
- [16]
Building on the platform, researchers are now applying these methodologies to map the brains of transparent vertebrate fish, larval zebrafish and adult Danionella, to model complex behaviour and neurodegenerative disorders.
- [17]
The original forecast of 500 specialists working for a decade amounts to about 5,000 person-years of work.
- [18]
The full central nervous system reconstruction covers about 6.6 times as many neurons as the 2020 hemibrain.
- [19]
The fly central nervous system contains roughly 550 times as many neurons as the C. elegans nervous system.
- [20]
Applying the claimed more-than-1,000-fold efficiency gain to that 5,000 person-year forecast implies roughly five person-years.
Sources
2 independent publishers whose own reporting we read for this story.
- dev.toเมื่อสมองแมลงวันกลายเป็นนักบิน AI: แผนผังที่เราไม่ได้ฝึกเลย
1 article · October 4, 2026
- neurosciencenews.comMicroscopy and AI Decipher Fruit Fly Neural Architecture
1 article · September 18, 2026
Topics and entities
Topics
- AI image segmentation
- Connectomics
- Electron Microscopy
- Neuroscience
Entities
- Howard Hughes Medical Institute
- HHMI Janelia Research Campus
- Google Research
- Caenorhabditis elegans
- MRC Laboratory of Molecular Biology
- University of Cambridge
- Drosophila melanogaster