[show-and-tell] {targets} and {condathis}: R pipelines Using R and Other CLI Tools

[c1au6i0]

In our bioinformatics lab, we love using targets for omics analysis. One challenge we faced was that many upstream steps in the analysis require other, non-R command-line interface (CLI) tools. To allow us to create targets reproducible pipelines that integrate both R and other CLI tools within a full R environment, my friend @luciorq created a CRAN package called condathis. This package was inspired by our experience with targets, and we wanted to share it here to show our appreciation for the work done on it. 👏👏

If you are curious, check out the condathis repository for more details.
Here’s also a practical example of a reproducible pipeline that employs targets and condathis to run common bioinformatic CLI tools for the in initial steps of an omics analysis.

[wlandau]

That's awesome, thanks for sharing condathis! FYI @joelnitta.

[joelnitta]

Cool! I had a similar idea to do this with docker containers.

[c1au6i0]

Thank you, will check that out @joelnitta :)

[luciorq]

Thank you, @c1au6i0 and @joelnitta,

I really enjoyed your rOpenSci presentation and the examples you shared, such as targets_vcf_example. They’re excellent illustrations of tackling a really common challenge in tool management.

With condathis, my primary goal was to adopt an “Environment as Code” approach for managing external tools while offering a familiar API akin to processx::run(). It also aims to improve portability, particularly for HPC systems that don’t support Docker or Singularity but can use Conda. Surprisingly, Conda works well beyond bioinformatics, especially given its strong support for Windows and macOS.

I had previously worked on an early version of dockerthis, which was intended to serve as a backend for condathis, enabling Conda environments to run within containers. However, after seeing your presentation, I realized that babelwhale addresses many of the features I had envisioned. Your contributed function, run_auto_mount(), especially caught my attention, as I had spent considerable time conceptualizing a similar API.

Ultimately, my ideal goal is to create a unified API (resembling processx::run()) that allows seamless execution of CLI tools across different backends—be it WSL2, Docker, Apptainer, Singularity, Podman, or others. However, supporting such a diverse range of backends involves significant effort.

Thank you again for the inspiration! Your work has given me a lot to think about.

[c1au6i0]

Great presentation, both of you! I just wanted to add one more thing about condathis. From a user point of view, a good thing is that it allows users to focus entirely on R without worrying about installing and setting up other tools, not even conda. Despite the name, the package automatically installs micromamba, which is self-contained and does not depend on Python.
Best,