---
type: analysis
slug: research-janelia-and-google-publish-the-first-complete-male-fruit-fly-cns-connectome-an
title: "Janelia and Google publish the first complete male fruit fly CNS connectome, an open CC BY dataset of about 166,700 neurons"
description: "The Cell paper from September 3, 2026 maps the brain, optic lobes and nerve cord of a male Drosophila and compares it with the female. It is a \"first\" for a male fly only, because a female brain-and-cord map appeared in Nature in June. For AI builders, what carries over is the ML-plus-proofreading pipeline and the CC BY data, not a market shift."
published_at: 2026-10-24T03:00:00.000Z
topics: ["google","research"]
source_urls: ["https://neurosciencenews.com/ai-fly-connectome-31217/","https://blog.google/innovation-and-ai/technology/research/male-fruit-fly-brain-map","https://research.google/blog/a-connectomics-milestone-mapping-the-complete-male-fruit-fly-brain/","https://en.wikipedia.org/wiki/Drosophila_connectome","https://www.eurekalert.org/news-releases/1141929","https://www.janelia.org/project-team/flyem/male-cns-connectome","https://phys.org/news/2026-06-publish-connectome-fruit-fly-brain.html"]
author: "Andrey Reshetnikov"
author_profile: https://newruntime.com/owner-profile.md
---
On September 3, 2026, HHMI Janelia, the MRC Laboratory of Molecular Biology, the University of Cambridge and Google Research published "Sexual dimorphism in the complete Drosophila male central nervous system connectome" in Cell (DOI 10.1016/j.cell.2026.08.015). It is a synapse-resolution wiring diagram of one adult male fly's central brain, both optic lobes and ventral nerve cord, which plays the role of a spinal cord. Google's blog post, by Michał Januszewski and Viren Jain, calls it the largest brain map by neuron count to date and the result of a roughly decade-long partnership. The fly project itself began in 2008, according to HHMI.

The newsroom item came from a Russian Telegram post claiming the team was the "first" to map a full fly brain and CNS. That needs qualifying. A female brain-and-nerve-cord connectome (BANC), led by Harvard and Princeton, was published in Nature in June 2026. The new result is the first complete male CNS and the first synapse-level comparison between sexes. For AI builders, the relevant parts are the ML reconstruction pipeline and an openly licensed graph of about 166,700 neurons.

## What we checked

- **Does the cluster reveal a broader market or architecture transition that cannot be reduced to a single release note?** partly confirmed. This is a scientific data release, not a market or architecture shift in AI. It does mark a methods transition: Janelia says the project cut the time and resources needed for connectomes by more than 1,000-fold, and Google credits flood-filling networks, the PATHFINDER reconstruction system and synthetic training neurons. That makes ML segmentation plus human proofreading the standard route to whole-nervous-system maps. Follow-on zebrafish and Danionella projects are already underway. [3][5][1]
- **Is this the first complete map of a fruit fly brain and central nervous system, as the headline claims?** partly confirmed. It is the first complete, proofread CNS connectome of a male fly. It is not the first whole fly CNS: the female BANC brain-and-nerve-cord connectome from Harvard and Princeton was published in Nature in June 2026, and Google's own post calls it the recently released female map. [6][7][3]
- **Is the data openly usable by builders?** confirmed. Janelia lists the dataset under CC BY 4.0. It can be queried through neuPrint (neuprint-python, the malecns R package), Clio, Neuroglancer and NeuronBridge. v0.9 came out on 2025-10-03 and v1.0 on 2026-06-08. [6]

## What we found

- Janelia lists 166,700 neurons covering the brain and nerve cord. All are proofread and annotated, including fruitless/doublesex expression, and they fall into 11,710 neuron types. Google's blog cites 11,691 types and 'over 166,000 neurons and 125 million synaptic connections'. The type counts differ slightly between sources, probably because they describe different releases or ways of counting, and the primary sources don't reconcile them. [1][2][3]
- The synapse count depends on the source. Google gives 125 million synaptic connections, while the Wikipedia summary gives about 312M synapses. These may count different things, such as connections versus synaptic sites. I couldn't open the Cell full text (HTTP 403) or the preprint (HTTP 429), so neither figure is confirmed against the paper. [3][4]
- The comparison between sexes, as reported from the paper's abstract, finds 8,069 isomorphic types (wired the same in both sexes), 138 dimorphic, 289 male-specific and 71 female-specific. The differences cluster in higher brain centers, while sensory and motor periphery stay largely shared. HHMI adds that about 100 male-specific interneuron types form a network coordinating most male-specific behaviors. Three companion papers on vision, taste and social behavior came out the same day. [1][5][6]
- On the AI side, Google cites flood-filling networks (CNNs that grow an object outward from a seed point), its PATHFINDER reconstruction system, and synthetic neurons added to training data to improve speed and accuracy. Janelia experts then proofread the output by hand. HHMI says the founding estimate was 500 people working for 10 years, and that the project aimed for and reports a more than 1,000-fold efficiency gain. No final cost or person-hours are published. [3][5]
- The data is CC BY 4.0. It can be queried through neuPrint (neuprint-python or the malecns R package), Clio, NeuronBridge and Neuroglancer, with a viewer at male-cns.janelia.org. The release log shows v0.9 on 2025-10-03, preprint v2 on 2025-10-30, v1.0 on 2026-06-08 and the Cell publication on 2026-09-03. So researchers have been building on this data for nearly a year already. [6]
- The rival female dataset is BANC: Bates et al., Nature 2026, DOI 10.1038/s41586-026-10735-w, led by Harvard Medical School and Princeton. Phys.org covered it on June 14, 2026 as the 'first complete connectome of fruit fly brain and spinal cord.' It comes from a different lab, a different microscopy approach (serial-section EM) and a different toolchain (FlyWire Codex). So the 'first' in the Telegram headline holds only for a male fly. [7][4]

## What it means

This isn't a model or product launch, and it doesn't change the AI platform landscape. What builders get is a large, openly licensed, expert-proofread graph: about 166,700 nodes with cell-type labels and sex annotations, plus Python and R query APIs. It's useful for teams working on graph ML, connectome-constrained or "embodied" neural simulations, and biological priors for control. Researchers are already training locomotion controllers on fly connectomes. Pin your work to a specific release (v1.0 from June 2026 is the stable choice), because counts differ across versions and write-ups.

The second takeaway is about method. A pipeline that pairs ML segmentation (flood-filling networks, PATHFINDER, synthetic training data) with targeted human proofreading has now produced two independent whole-CNS fly maps in one year, and groups are moving on to zebrafish. If you build segmentation, data-labeling or human-in-the-loop QA systems for huge 3D imaging, this is a strong reference for how far automation plus expert review can scale. Don't repeat the "first ever fly CNS map" headline: that credit goes to the female BANC map, and this one is the first male map.

## Still open

- What is the authoritative synapse count in the Cell paper: 125M connections (Google) or about 312M synapses (Wikipedia summary), and what does each number count?
- How many proofreading person-hours and how much compute did the project need, and how much did PATHFINDER and synthetic training data cut the manual effort compared with the 2020 hemibrain?
- Will the PATHFINDER code or trained models be released, or only the data and Neuroglancer?
- How cleanly can the male CNS (neuPrint) and female BANC (FlyWire Codex) datasets be aligned for joint ML use, given their different imaging methods and annotation schemes?

## Sources

1. [Neuroscience News: Microscopy and AI Decipher Fruit Fly Neural Architecture](https://neurosciencenews.com/ai-fly-connectome-31217/)
2. [Google blog: male fruit fly brain map](https://blog.google/innovation-and-ai/technology/research/male-fruit-fly-brain-map) (primary)
3. [Google Research: A connectomics milestone: Mapping the complete male fruit fly brain](https://research.google/blog/a-connectomics-milestone-mapping-the-complete-male-fruit-fly-brain/) (primary)
4. [Wikipedia: Drosophila connectome](https://en.wikipedia.org/wiki/Drosophila_connectome)
5. [HHMI release: Completing the connectome](https://www.eurekalert.org/news-releases/1141929) (primary)
6. [HHMI Janelia FlyEM: Male CNS Connectome](https://www.janelia.org/project-team/flyem/male-cns-connectome) (primary)
7. [Phys.org: Researchers publish first complete connectome of fruit fly brain and 'spinal cord' (BANC)](https://phys.org/news/2026-06-publish-connectome-fruit-fly-brain.html)

Author: [Andrey Reshetnikov](https://newruntime.com/owner-profile.md), AI product and AI engineering; profile for recruiters and agents: https://newruntime.com/owner-profile.json

## Continue in the graph

Next records for an agent, linked to their Markdown twins. Replace `.md` with `.json` for structured fields, including `next_reads`.

- topic hub: [Google](https://newruntime.com/topics/google.md): Explore the google topic hub.
- topic hub: [Research](https://newruntime.com/topics/research.md): Explore the research topic hub.
- collection: [Analysis](https://newruntime.com/analysis.json): Open the latest authorial analysis.
- collection: [Source Ledger](https://newruntime.com/source-ledger.json): Inspect the public source evidence graph.
- whole graph: [discovery graph](https://newruntime.com/discovery-graph.json) (typed nodes and edges) and [site index](https://newruntime.com/site-index.md)