snakemake_hippocampal_unfolding

This is a pipeline for visualizing output of hippocampal autotop pipeline by Jordan Dekraker.

Example output

Quality control gif images:

Unfolded Maps plotted (thickness):

Violin Group plot (thickness):

Lineplot Group plot (thickness):

Other output include

• npz files for each subject containing unfolded maps
• pkl and csv file containing flattened unfolded maps in convient format for comparison

Requirements

1. Snakemake installation instructions or if using compute canada

2. Gif your nifti singularity container --> https://hub.docker.com/r/kaitj/gif_your_nifti

   singularity pull docker://kaitj/gif_your_nifti

3. Python packages

• pandas
• numpy
• matplotlib
• seaborn
• scipy
• sci-kit image
• sci-kit learn

Usage

1. Populate participant_id.tsv file with subjects to be run in pipeline
2. Populate config.yaml file with appropriate features for your dataset:

config.yaml

participant_tsv: participant_id.tsv
input_dir: /home/myousif9/scratch/hcp_unfolding3_jdkrek                                     <-- top level directory for your dataset
boundary: /scripts/BigBrain_ManualSubfieldsUnfolded.mat   <-- do not change this
singularity: /scratch/myousif9/singularity/gif_your_nifit_miykael.sif                       <-- point this to where ever this singularity container is located

coords:
  - AP
  - IO
  - PD

features:         <-- name of features in mat files (add to this if you have more features to unfold)
  - GI
  - streamlengths
  - qMap
  
unfold_mat:       <-- name of mat files without extension (add to this if there is another mat file to unfold)
  - surf

hemi:             <-- list of hemispheres to visualize (add R or L or both R and L)
  - L

cmap: plasma      <-- choose the color map you would like for your unfolded maps 

link to list of colormaps

3. Ready to run the pipeline now!

• Run with by using command snakemake or which ever way works for your system or compute cluster etc.
• To generate report by running snakemake --report

Acknowledgements

Thanks to Roy Haast, Jason Kai and Jordan Dekraker for providing code and help in making this pipeline.