NeuroConstellation: Interactive Visualisation and Analysis of Million-Neuron Datasets Using Game Engine Technology

NeuroConstellation visualises and analyses large neuron-level simulation datasets in real time. It runs on Windows with an NVIDIA GPU and uses Unity 3D. The largest dataset tested is the 5.28-million-neuron human CA1 model, which runs at 60-100 FPS on a consumer card. This deposit contains: - NeuroConstellation Software.zip: the Windows application. - NeuroConstellation_Converter_v1.0.zip: a Python GUI converter that reads NWB, Phy, Kilosort, and CSV files into the HDF5 format the application uses. The subfolder experimental_data_examples/ inside this archive contains the scripts that reproduce Figure 11 of the paper. - activity_network_mouse.hdf5 and activity_network_human.hdf5: two sample datasets so you can try the application without running the converter first. - Validation.zip: scripts that check the GPU output against a NumPy reference implementation (used to produce Figure 2 of the paper). - USER_GUIDE.md and NeuroConstellation_User_Guide.docx: installation, controls, data format, validation workflow, and troubleshooting. - README.md and LICENSE. All code is released under the MIT licence. If you use NeuroConstellation in your research, please cite the companion paper: Mirani S.A., Memon P., Enrico P., Urgese G., Solinas S.M.G. NeuroConstellation: Interactive Visualisation and Analysis of Million-Neuron Datasets Using Game Engine Technology. iScience, 2026.