This repository contains research code and figures of the paper Simulation-based inference for efficient identification of generative models in connectomics. In the paper, we show how to use simulation-based inference (SBI) to infer parameters of computational models used in connectomics.
The repository is based on a small python package called consbi
that allows to simulate different wiring rules in the structural model of the rat barrel cortex.
Additionally, it contains jupyter notebooks with tutorials
and code for reproducing the figures
shown in the paper.
Binary files with data
and results
for the work presented in the paper are stored using git-lfs
.
The repository for running SBI including detailed tutorials is located at https://github.com/mackelab/sbi.
Please reach out and create an issue if you have any questions or encounter problems with using this repository.
Tutorials for the SBI workflow and code for reproducing the figures are available as Jupyter Notebooks that can be opened in the browser (without executing them).
To run and play around with the code, you need to clone and install this repository locally, e.g., in the command line, run:
git clone https://github.com/mackelab/sbi-for-connectomics.git
cd sbi-for-connectomics
pip install -e .
If you then start an jupyter notebook
server locally you should be able to open and execute all notebooks.
The repository contains a number of large files that are provided via git-lfs
.
See the git-lfs
documentation for installation and general info, https://git-lfs.com.
Once git-lfs
is installed locally, one can pull the large files contained in the repository just by cloning or pulling it.