cvetkovicdamjan/neurilium

Neurilium is a small program that simulates the electrical activity of neurons and their connections, rendered visually in real time. The primary example in the README is the fruit fly brain, which is currently the most complete whole-brain neural map available to researchers. You feed it real connectome data (the wiring diagram of a brain) and watch the simulated neurons fire. Getting started means downloading three CSV files from the FlyWire project's public dataset: one listing neurons, one listing connections between them, and one with spatial coordinates. You place those files in a data folder, clone the repository, and run the program with a single Cargo command. Navigation in the 3D view uses mouse and keyboard controls described in the README. The simulation uses a model called Izhikevich neurons, which the README explains are simple to compute while still producing realistic neuron firing behavior. The GPU work runs through wgpu, a Rust library that abstracts over graphics and compute APIs. The author notes this combination replaces the more traditional C++ plus CUDA plus Vulkan stack in a more unified way. The codebase is intentionally small: around seven files handling loading, rendering, camera, and simulation. The README also points to other publicly available brain datasets you could adapt the code to simulate, including a small region of human brain tissue, a mouse cortex sample, a zebrafish whole-brain reconstruction, and the full nervous system of a roundworm. No dataset download links are provided for those, but the README notes they can be found with some searching. This project is open to experimentation. The README frames it as code that is easy to modify rather than a finished product, and acknowledges the FlyWire team for making the fruit fly data publicly available.