Neutralisation analysis launcher

Launches plaque neutralisation analyses when new data is exported.

• Monitors a given directory for new experiments in a cron job.
• When newly exported data is detected, submit analysis tasks to a job queue.

File system monitoring

1. A cron job is used to list all the files matching a regex pattern in a given CAMP directory.
2. Store a hash of the ordered filenames and the filenames themselves.
3. The next time the cron job is run, compare the hash to the previous value, if it is different then identify new filenames.

Job queue

The job queue is made with Celery and redis. There are separate queues for IC50 calculations and image stitching tasks. IC50 calculations are performed via the plaque_assay library. Image stitching is done with a scikit-image scripts and saved to a directory on CAMP.

Requirements

This requires an installation of redis-server, celery and a MySQL driver.

• Redis 6.0.10
• Celery 4.4.7
• Celery flower (optional)
• miniconda3 (or other python installation with package manager)
• plaque_assay

To run:

To start all the celery workers:

# starts all processes in a tmux sesssion
./launch.sh

Then you'll need to create a cronjob for the snapshotter, which will detect new exports and send them to the celery workers. An example of the cronjobs are listed below:

# run neutralisation snapshot every 5 minutes to detect new files
*/5 * * * * source $HOME/.bashrc; $HOME/miniconda3/bin/python3.8 $HOME/launcher/launcher/run.py
*/5 * * * * source $HOME/.bashrc; $HOME/miniconda3/bin/python3.8 $HOME/launcher/launcher/run_titration.py

---

Drag and drop portal

There is also a "drag-and-drop" version of the neutralisation analysis, which enables running the analysis data exported to a specific directory, but saving the output to files rather than uploading to the LIMS database. This is useful for troubleshooting assays or other experiments which don't fit into the typical neutralisation pipeline.

TODO: more info, where, how to set up.

Dilution swapping

Sometimes during the assay, human error means dilutions are in the wrong quadrants. There is a "dilution-swap-portal" which acts on the indexfiles and can correct the positions of the dilutions to enable the indexfiles to then be used in the normal plaque_assay analysis.

TODO: more info, where, how to set up.