Blazed oblique plane microscopy reveals scale-invariant inference of brain-wide population activity

The notebooks folder contains a jupyter notebook that downloads all calcium-dependent fluorescence traces used in the paper and generates the data panels of figures 3 to 5.

Colab

The notebook contains code that sets up a suitable environment in Google Colab (click here: ). In case there is a warning about conflicting cupy versions, it can be ignored.

A Colab Pro pro instance with a A100 GPU is needed to compute the pairwise correlations in Figure 3. A free Colab GPU instance is sufficient to perform the analysis of the remaining figures 4 and 5.

To install software and download the data locally

Using mamba (or conda) as package manager:

 mamba create -n hoffmannetal -c rapidsai -c conda-forge -c nvidia rapids=23.10 python=3.10 cudatoolkit=11.8
 mamba activate hoffmannetal 
 git clone https://github.com/danionella/hoffmann_et_al_2023.git
 pip install --ignore-installed  --quiet ./hoffmann_et_al_2023
 wget "https://gin.g-node.org/danionella/Hoffmann_et_al_2023/raw/5a3146dc108208415f87bf17ebce37d566b28208/20230611_export_3.h5" -O data.h5

A faster but non permanent data repository:

wget "https://owncloud-ext.charite.de/owncloud/index.php/s/H97Qi8haRYLZu4e/download" -O data.h5

Tested on:

Intel(R) Core(TM) i7-7820X CPU @ 3.60GHz
Ubuntu 20.04.6 LTS
125 GB RAM
NVIDIA GeForce RTX 3090 24GB

Hardware requirements: For Figure 3: GPU with memory >= 24 GB For Figure 4 and 5: GPU with memory >= 16 GB