Tomography Flow Example

This repository demonstrates how Tomopy reconstructions can be automated using Globus Flows. The notebook aims to show how an individual flow can be used to integrate different light sources and compute facilities.

Configuring a compute environment

We use Globus Compute to perform Tomopy remotely. To plug your own Compute endpoint in you will need to install globus-compute-endpoint and the tomopy dependencies listed in compute-requirements.txt

Repository Structure

tomography_flow/ ├── recipes/ │ └── Dockerfile # OCI recipe for building the container image ├── tomo_flow/ # Package containing the tomo_flow module (with recon.py, etc.) │ └── recon.py │ └── run_tomopy.py # Script to run the tomographic reconstruction workflow │ └── flow_definition.json ├── notebooks/ │ └── tomo_nsls_244.ipynb │ └── Tomography-flow.ipynb ├── imgs/ └── README.md # This file

Building the container Image

cd recipes
podman build -t tomography_flow:latest .
cd ..

Running the container

podman run -it --rm \
-v "$PWD/NSLS2":/data \
-v "$PWD/tomography_flow":/app \
tomography_flow:latest python3 /app/tomo_flow/run_tomopy.py \
  --proj-file /data/scan_00244/proj_00000.hdf \
  --angles-file /data/scan_00244/scan_00244.nxs \
  --dark-file /data/scan_00245/dark_00000.hdf \
  --flat-file /data/scan_00245/flat_00000.hdf \
  --recon-init 1000 \
  --recon-end 1010 \
  --output-dir /data/recon_output \
  --show-images