Pipeline and analysis tools for nuclear phenotyping

This repository is designed for analysing images of nuclei using deep learning However, it is specialised for analysing 2D max-Z projections of organoids, though it may have general usage outside of this specific task.

This does contains a full snakemake pipeline to:

  • Train UNet and *Dist models
  • Run inference using models on Cellesce data
  • Convert inference images to per-nuclei features
  • Compile nuclei into a csv

Add secrets

set -o allexport; source secrets.env;set +o allexport

Install env

make install.snakemake.env

Test Snakemake

snakemake --dry-run

Produce graphs

python splines.py

TODO:

  • Get automatic zenodo uploading working
  • Seperate the UNet package into it's own git repo
  • Add Figures