graph analysis with network x

This script was designed to calculate different single-cell spatial statistical metrics for fluorescence FFPE image data. The cells should have a class, centroid coordinates and belong to a certain image and region of interest (ROI). It can potentially be used in other applications where class and coordinates are available for objects belonging to an image and ROI.

dependencies: pandas numpy matplotlib libpysal networkx seaborn scipy

new dependencies: alive-progress

optional dependencies: geopandas

how to run:

python nwx_analyze.py -h

example

NOTE: --tiff_dir is not used in this version, meaning the image directory prompt is useless

python nwx_analyze.py -i ~/Documents/Ovarial_22/ML_TMA1/obj_class_TMA1.csv -p CD45:PANCK --tiff_dir ~/Documents/Ovarial_22/imj_out_1

Notes

Regarding class1:class2

while developing the script, class1 was assumed to be immune cells and class2 cancer cells, this means for example that group degree centrality is calculated for whatever is on the left side!

QUICKSTART

• Make sure you have conda!

• Output the results of your cell detections in QuPath to the default format. The column numbers will be treated as:

  0: image_name, 1: cell_class, 2: name, 3: ROI_name, 4: x-coordinate, 5: Y-coordinate

• Open the output tsv to determine what decimal point QuPath was using (if it is not . you need to specify)

• Change any UTF-8 incompatible characters in the column titles of the output tsv to compatible ones to avoid crashes.

• Create new conda environment with all dependencies:

conda create -n graph_analysis -c conda-forge -c bioconda pandas numpy matplotlib libpysal networkx seaborn scipy geopandas alive-progress

• Activate your environment (start here if you already set the environment up) conda activate graph_analysis

• Clone this repo and navigate to the local directory with the conda prompt

• Get the help text for constructing your CLI command

python nwx_analyze.py --help

Troubleshooting:

If you get an error that seems to indicate a package is missing, first check that your conda environment is properly set up and that you have activated it - then if the problem persists write an issue on github! QuPath oftentimes writes out UTF-8 incompatible characters in the column titles of tsv-files. This could be a source for failed execution of the script. Change titles to only contain compatible characters and try again.