Here I've included a framework for running BIRDMAn on a SLURM server.
1) Install BIRDMAn. If you have trouble with installation here, feel free to use the .yml file that I've provided in the repo:
conda env create -f birdman.yml
2) Clone this repo.
git clone https://github.com/ahdilmore/MARS_Birdman.git
3) Create your model. Templates are provided in birdman/src/model_amyloid_single.py and birdman/src/model_apoe4_single.py.
4) Modify chunking file. Templates are provided in birdman/src/amyloid_birdman_chunked.py and birdman/src/apoe4_birdman_chunked.py
4) Modify overall script, then run it. Templates are provided in birdman/amyloid/zebra.sh and birdman/apoe4/zebra.sh. You will know that the script is complete when (1) SLURM emails you or (2) .nc files are created in your inferences directory for each feature.
sbatch zebra.sh
5) Modify path in summarize-inferences, then run it. You will know that the script is complete when (1) SLURM emails you or (2) You have a .tsv file in your inferences-results directory.
sbatch summarize-inferences.sh
6) Import the .tsv files into the analysis notebook to analyze results.