Differentiable Biorealistic V1 Model Resources
These resources accompany our bioRxiv manuscript, Deep-learning-assisted simulation of a cortical circuit (Ito et al. bioRxiv 2026).
The project introduces a biologically grounded mouse V1 model that combines structural connectivity, cell-type-specific physiology, and TensorFlow-based training and evaluation workflows. This page points to the main repositories and supporting data resources.
Related Resources
Add "Atom - 106 Item" to this slot. This text won't show when published.
Main repo: biorealistic-v1-model
Network build and PointNet-based simulation workflow for the biorealistic V1 model.
TensorFlow-based differentiable simulator: V1_GLIF_model
TensorFlow implementation for training and evaluating the GLIF-based V1 model.
Digital twin analysis repo: biorealistic-v1-model-digital-twin
Analysis pipeline for digital-twin training, MEI generation, and spatial-frequency analysis.
Add "Atom - 106 Item" to this slot. This text won't show when published.
EM analysis repo: biorealistic-v1-model-em-analysis
Electron-microscopy-based analysis used to parameterize synaptic connectivity.
Functional-property table repo: v1_functional_property
Neuropixels-derived functional-property tables used as analysis targets and comparisons.
Prebuilt SONATA network download
Optional prebuilt network files for the TensorFlow workflow.