Modern bioinformatics pipelines can be incredibly complex, but all tend to follow a common pattern: they start with raw data, then pass the data through various programs until arriving at a final result. If this is done in an ad-hoc, unorganized fashion, the results may never be reproducible or even worse, unreliable and/or wrong. Pipeline management software is therefore essential to obtain results that are robust and reproducible. The targets R package is a recently developed workflow manager that comes with many excellent features for bioinformatics, including data caching, pipeline-level parallelization, and HPC support.

In this hands-on workshop, I will demonstrate how targets can be used in concert with other tools like docker and conda to orchestrate modular, reproducible bioinformatics pipelines. The workshop will feature variant-calling as an example, but the concepts and tools can be applied to nearly any analysis.

Pre-requisites: Basic familiarity with R. Installations of recent versions of R, conda, and docker.

Duration: 2 hours