Running Pipelines in MagellanMapper#

TODO: update for new pipelines

Start a processing pipeline#

Automated processing will attempt to scale based on your system resources but may require some manual intervention. This pipeline has been tested on a Macbook Pro laptop and AWS EC2 Linux (RHEL and Amazon Linux based) instances.

Optional dependencies:

  • ImageJ/Fiji with the BigStitcher plugin: required for tile stitching; downloaded automatically onto a server when running deploy.sh

  • ImageMagick: required for exporting a stack of planes to an animated GIF file

  • FFMpeg: required to export a stack to a movie format such as MP4

  • Slack incoming webhook: to notify when tile stitching alignment is ready for verification and pipeline has completed

Local#

Run a pipeline in pipelines.sh.

For example, load a .czi file and display in the GUI, which will import the file into a Numpy format for faster future loading:

bin/pipelines.sh -i data/HugeImage.czi

To sitch a multi-tile image and perform cell detection on the entire image, which will load BigStitcher in ImageJ/Fiji for tile stitching:

bin/pipelines.sh -i data/HugeImage.czi -p full

See bin/pipelines.sh for additional sample commands for common scenarios, such as cell detection on a small region of interest. The file can be edited directly to load the same image, for example.

Server#

You can launch a standard server, deploy MagellanMapper code, and run a pipeline. See tools for AWS cloud management for more details.