Terramind Labs / The fly brain / Lab
Our paper Code Built in Nucleus

Poke a fly's brain.

Every dot is one neuron of a real fruit fly, wired the way that fly was wired. Give her something to sense and watch the signal cross her brain.

resting
Our drawing. Only the model's neurons move it.
Feeding neuron0
Grooming neurons0
Acceptance neurons0
Other neurons responding to the song0
Loading the brain

Courtship, honestly

A female fly signals that she is willing to mate by opening her vaginal plates, and she does it in response to a male's courtship song. A pair of female-specific neurons called vpoDN controls it. They get excitatory input from song-tuned auditory neurons (vpoEN) and from pC1 neurons, which encode her mating status (Wang et al., Nature 2021).

The model was built and checked for feeding and antennal grooming, not for courtship, and it has no notion of what a song sounds like. Switching pC1 on is our way of probing the circuit. It is not a claim about what makes a real fly receptive.

The model's authors did not run these tests. We did (10 runs per setting); they are not peer-reviewed. Source: doi.org/10.1038/s41586-020-2972-7

What this is, and what it isn't

This is the computer model from Shiu et al., Nature 2024, running in your browser on the published wiring map of one adult female fruit fly. The page uses version 783 of the map as shipped with the model code (138,639 neurons). The paper itself used an earlier version (127,400 neurons).

Our port of the model is our own work and is not peer-reviewed. We checked it against the authors' reference code: identical spike for spike on a test network, and firing rates within about 2 Hz of theirs for the sugar, water, sugar-with-bitter and antenna actions (10-run averages, on the map version the paper used).

Paper: doi.org/10.1038/s41586-024-07763-9 · Model code: github.com/philshiu/Drosophila_brain_model

Sources and credits

The model. Shiu, P.K. et al. (2024). A Drosophila computational brain model reveals sensorimotor processing. Nature 634, 210–219. doi.org/10.1038/s41586-024-07763-9. The engine on this page is our port of the authors' code (github.com/philshiu/Drosophila_brain_model, MIT licence, © 2023 Philip Shiu and Nico Spiller). The sugar, water, bitter and antenna neuron lists come from that repository.

The wiring map. The FlyWire connectome of an adult female fruit fly brain, version 783, in the form distributed with the model code. Dorkenwald, S. et al. (2024). Neuronal wiring diagram of an adult brain. Nature 634, 124–138. doi.org/10.1038/s41586-024-07558-y. FlyWire releases its data under CC BY-NC 4.0 (flywire.ai/guidelines). This page is non-commercial.

Cell types and neuron positions. Schlegel, P. et al. (2024). Whole-brain annotation and multi-connectome cell typing of Drosophila. Nature 634, 139–152. doi.org/10.1038/s41586-024-07686-5. Annotation table: github.com/flyconnectome/flywire_annotations, which lists every paper to cite.

Real-fly results in the Bet game. Supplementary Table 3 of Shiu et al. (2024).

Courtship circuit. Wang, K. et al. (2021). Neural circuit mechanisms of sexual receptivity in Drosophila females. Nature 589, 577–581. doi.org/10.1038/s41586-020-2972-7.

Our own work. The tests behind this page are written up in our preprint, "What the Fly's Wiring Map Predicts Without a Simulation" (not peer-reviewed): read the PDF. The plans, code and result files are public: github.com/terramind-org/science. Corrections are listed on the project page.

Others got here first. People have run fly brain simulations in a browser before us, including fly-explorer by Brandon Cho and webgpu-fly by Ahmet Barış Günaydın.

The fly drawing and this page are by Terramind. Our port of the model is not peer-reviewed.