Build With a Fruit Fly Brain 🪰

A connectome is a wiring map: neurons are the nodes, and synaptic connections are the links between them. You can explore the structure and build analysis tools around it. A wiring map alone does not capture every property or moment-to-moment state of a living brain.

• TRY IT IN YOUR BROWSER — FlyWire Codex lets you search neurons, inspect connections and view cells in 3D. Its interactive apps require Google sign-in. Start here: codex.flywire.ai/
• TAKE A GUIDED TOUR — FlyWire Academy provides lessons and activities. Start with a small set of cells, then inspect their incoming and outgoing connections: codex.flywire.ai/academy_home?dataset=fafb
• TRY EXAMPLE SEARCHES — Use the official sample queries rather than guessing the search syntax: codex.flywire.ai/sample_queries?dataset=fafb
• CHOOSE A DATASET — The official directory currently lists FAFB v783 (female adult brain), BANC v888 (female brain and nerve cord), MANC v1.2.1 (male nerve cord), MAOL v1.1 (male right optic lobe) and MCNS v1.0 (male brain and nerve cord). These cover different anatomy and specimens. Record the selected release and export filters; do not mix neuron IDs or compare counts without checking scope. Directory checked 7 October 2026: codex.flywire.ai/
• GET DATA AND CODE — The Seung Lab’s FlyConnectome repository provides notebooks for querying connections and annotations through CAVE and accessing image volumes and meshes. It also points to bulk data downloads through Codex. Start with a small export: github.com/seung-lab/FlyConnectome
• PYTHON TOOLS — fafbseg provides FlyWire-specific tools that work with navis. Follow the current documentation for authentication, supported releases and installation: github.com/navis-org/fafbseg-py
• OPEN ANNOTATIONS — The FlyWire annotations repository supplies neuron labels and related data for the v783 release. Check its release notes and citation requirements: github.com/flyconnectome/flywire_annotations
• EXPLORE A PUBLISHED SIMULATION — Shiu and colleagues’ model uses connectome-derived connectivity and neurotransmitter identity to study sensorimotor processing. The authors’ code and setup instructions are here: github.com/philshiu/Drosophila_brain_model. The example notebook is github.com/philshiu/Drosophila_brain_model/blob/main/example.ipynb . The default configuration uses v630; follow the README’s separate configuration for v783. Read the research alongside it: pmc.ncbi.nlm.nih.gov/articles/PMC11446845/
• FIVE PROJECT IDEAS — 1) A neuron explorer with a table and a small connection graph. 2) A dashboard of cell-type counts. 3) A pathway finder between selected neurons. 4) A comparison of connection patterns across brain regions. 5) A small circuit simulation with explicit assumptions. The first four analyse structure; the fifth adds a model of activity.
• FIRST BUILD: A NEURON EXPLORER — 1) Pick one dataset release and a small cell group in Codex. 2) Export its neuron labels and connections and record the source, release and filters. 3) Inspect the actual column names; keep neuron IDs as strings. 4) Use Python to join labels to connections and sum synapse counts for each neuron pair. 5) Display a searchable table and a graph of a few dozen cells. 6) Check several edges against Codex and label omitted connections or thresholds. Ask the prompt below for starter code matched to your export.
• WHAT THE DEMO DOES NOT ESTABLISH — A connectome-based simulation adds assumptions about neuron dynamics and inputs. Its outputs are model predictions, not evidence that a living brain was literally uploaded. The published model studies specific biological behaviours; it does not establish that a fly understands markets or can trade profitably.