You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

MaleCNS v1.0 — Male Fruit Fly Central Nervous System Connectome

Dataset Description

This repository mirrors the MaleCNS v1.0 connectome — the first complete wiring diagram of the central nervous system (brain + ventral nerve cord) of an adult male Drosophila melanogaster (fruit fly).

The original dataset was produced by a collaboration between:

  • HHMI Janelia (FlyEM Project Team)
  • University of Cambridge (Dept. of Zoology)
  • MRC Laboratory of Molecular Biology
  • Google Research

Original source: male-cns.janelia.org/download

License: CC-BY 4.0 — original license preserved.


What's in this repo

The flat connectome files from the Janelia GCS bucket (gs://flyem-male-cns/v1.0/connectome-data/flat-connectome/), including:

  • Connectivity data — neuron-to-neuron synapse connections (from neuron, to neuron, connection weight)
  • Neuron annotations — cell type labels, neurotransmitter predictions, soma locations
  • Skeleton files — 3D morphology of individual neurons in SWC format

Dataset Statistics

Property Value
Neurons ~166,700
Synaptic connections ~25 million
Coverage Full CNS (brain + ventral nerve cord)
Resolution 8nm isotropic EM
Release version v1.0 (June 8, 2026)

Intended Use

This mirror is being used as the data backbone for a personal AI assistant project (Jarvis) that:

  1. Runs the connectome as a live spiking neural network simulation (via Brian2)
  2. Feeds the simulation encoded representations of real task logs
  3. Applies Hebbian learning and neurogenesis rules on top of the static connectome
  4. Uses regions of network growth as signals for capability gaps in the AI system

This is an experimental research project. The connectome is not being modified — growth rules are applied in simulation only, on top of the original graph.


How the Simplified Activity Model Works

The simulation uses leaky integrate-and-fire (LIF) neurons — a simplified but biologically grounded model where:

  • Each neuron accumulates charge from incoming synaptic signals
  • When charge crosses a threshold, the neuron fires (emits a spike)
  • Charge decays over time if threshold isn't reached

On top of the static MaleCNS topology, the following rules are applied:

  • Hebbian strengthening — synaptic weights increase between neurons that fire together repeatedly
  • Synaptic pruning — connections that never activate across N simulation steps are weakened and removed
  • Neurogenesis trigger — when a neuron cluster consistently saturates (hits its firing ceiling), new nodes are spawned and attached to that region

Citation

If you use this data, please cite the original MaleCNS paper:

Schlegel et al. (2024). Whole-brain annotation and multi-connectome cell typing quantifies circuit stereotypy in Drosophila. bioRxiv.

And link back to the original source: male-cns.janelia.org


Acknowledgements

All connectome data belongs to the original authors and is reproduced here under CC-BY 4.0. This repository does not claim ownership of the underlying biological data.

Downloads last month
307