Magellon demo

Install the Dockerized version of Magellon

1. Install Docker

   1. For Linux distributions where you don't have root, a system administrator will need to install Docker and make you a docker user

      1. Additionally, the Nvidia conatiner toolkit ("nvidia-toolkit-container") must be installed and configured in order for GPU operations to work

      2. Users can check for installation by executing "which nvidia-ctk"

   2. For Mac OS and Windows Docker Desktop is preferred https://docs.docker.com/desktop/

2. Download the demo data

   1. Download using Globus. Data can be found here: https://app.globus.org/file-manager?origin_id=4ad48aa4-a13a-4a8e-9a2b-da29eafbd3f9&origin_path=%2F

3. Download the Magellon source code https://github.com/sstagg/Magellon (e.g. git clone https://github.com/sstagg/Magellon.git)

4. Create a Magellon directory to house processing files.

   1. e.g. /scratch/users/sstagg/magellon

5. Determine what CUDA version is installed on your system.

   1. This is required for motion correction. On systems without a GPU, motion correction will be skipped.

   2. e.g. execute "nvidia-smi" and record the CUDA version

6. In a terminal, navigate to the Docker subdirectory in the Magellon source (e.g. "cd /Users/sstagg/Documents/GitHub/Magellon/Docker")

7. In a terminal execute the installation script, also providing the Magellon processing directory and your CUDA version

   1. "bash setup-magellon.sh <magellon processing directory> <CUDA version>

   2. e.g. "bash setup-magellon.sh /scratch/users/sstagg/magellon 11.8"

   3. This installs all the Magellon containers, sets up the required environment file ".../Magellon/Docker/.env", and sets up the Magellon home directory stucture

8. After the installation script completes, it will launch several webpages including the Swagger interface (http://localhost:8000) and the Magellon frontend (http://localhost:8080/en/panel/images)

9. Import data

   1. move the data you downloaded from Globus into the Magellon GPFS directory (e.g. /scratch/users/sstagg/magellon/gpfs)

      1. This allows the Magellon interface to be aware of the data. All data imported into Magellon should be located in this directory

   2. In the Magellon frontend (http://localhost:8080/en/panel/images), click the three stacked lines icon in the top left

   3. Click Import and select the Magellon import tab

      1. If you have correctly moved the demo data into the Magellon GPFS directory, you will see the session name in the GUI

      2. Double click the 24dec03a directory icon

      3. Then double click the session.json file

      4. Finally click the IMPORT DATA button

         1. 
            

            

            
            

10. This data will now start importing. This should take 5-60 minutes or more depending on the computer you're using.

    1. If you are running on a Linux machine with a compatible GPU, Magellon will align frames which is why the times vary so much.

11. Go to the Magellon frontend (http://localhost:8080/en/panel/images)

    1. Select 24DEC03A from the "Session" pulldown

    2. Explore!

NOTE: If you are installing on Linux, some of the files may be owned by the Docker user. If you want to try installing again using the installation script, you will need to wipe out the magellon processing directories. A script is provided in the Docker directory that will wipe the directories except for the gpfs directory. The script can be run by executing "bash cleanup.sh <magellon processing directory>". WARNING - this will delete all the magellon subdirectories in that directory, so use at your own risk.