Looper is a project manager written in python; it simplifies submitting pipeline jobs for your samples either on a cluster or on your local computer.
Once you’ve built a pipeline (for example, using pypiper), you need a way to deploy that pipeline across lots of samples. Looper helps you do that.
You use a
yaml file to describe your project. It points to a
csv listing each sample you want to run, and has fields for data input location, pipeline output location, and other project-specific variables. Looper reads this metadata and submits pipeline runs for each sample.
Looper makes it easy to:
- only submit jobs that haven't already been submitted
- run multiple pipelines on each sample
- interface with any kind of pipeline
- collate inputs from different locations on disk
- request different resources for different input file sizes
- monitor which jobs are running or failed
- run different pipelines on different types of data
- use sample objects for downstream (post-pipeline) data analysis
The best place to start is the Documentation, which includes tutorials, installation instructions, and more.