This repository contains data, code, and results for the manuscript "Comparing spatial null models for brain maps" by Ross Markello & Bratislav Misic (NeuroImage, 2021). We investigated how well different null model implementations account for spatial autocorrelation in statistical analyses of whole-brain neuroimaging data.

We've tried to document the various aspects of this repository with a whole bunch of README files, so feel free to jump around and check things out.

Itching to just run the analyses? You'll need to make sure you have installed the appropriate software packages, have access to the HCP, and have downloaded the appropriate data files (check out our walkthrough for more details!). Once you've done that, you can get going with the following:

git clone https://github.com/netneurolab/markello_spatialnulls
cd markello_spatialnulls
conda env create -f environment.yml
conda activate markello_spatialnulls
pip install parspin/
make all

If you don't want to deal with the hassle of creating a new Python environment, download the Singularity image that we used to run our analyses and run things in there:

git clone https://github.com/netneurolab/markello_spatialnulls
cd markello_spatialnulls
wget -O container/markello_spatialnulls.simg https://osf.io/za7fn/download
singularity run container/markello_spatialnulls.simg make all

If you want a step-by-step through all the methods + analyses, take a look at our walkthrough.

Open an issue on this repository and someone will try and get back to you as soon as possible!