The framework is built to allow any number of pipelines. They are implemented using Nextflow.io, and take a sample metadata file as input. Normally the pipelines are launched by the executor, upon detection up a new analysis.
In the standard offer, three types of analyses are offered: - basepipe: performs mapping, variant calling and QC - triopipe: performs join calling - annopipe: performs annotation and reporting
For a single sample, two analysis folders and metadata would have been created: one basepipe and one annopipe (which depends on the genepanel). For a trio, five analysis folders and metadata would have been created: three basepipe (one for each of the members), one triopipe (for joining them) and one annopipe.
You can run a pipeline manually by running:
SETTINGS=/path/to/settings.json exe/pipeline/basepipe --analysis /path/to/my_analysis.analysis
How to create new pipeline
Since an analysis in the system consists of a single entry point (command), your pipeline must run the whole analysis from start to end in one step.
mypipe.nf: a nextflow pipeline
mypipe-dependency: python script for checking dependencies
mypipe-post: python script for calling Quality Control and delivering files
- a symlink
../lib for utility classes
exe/pipeline create an entry point script
No modification of the executor is necessary.