Skip to content

Latest commit

 

History

History
53 lines (43 loc) · 2.6 KB

contributing.md

File metadata and controls

53 lines (43 loc) · 2.6 KB
Raw

Contributing to tidyomics

Making contributions to existing tidyomics packages

  1. Check the open challenges page, or ask one of the community members about adding a new open challenge
  2. Reach out to a package developer, either through a GitHub Issue, or on the Bioconductor community Slack, #tidiness_in_bioc channel
  3. Discuss building a pull request with current members of the tidyomics development team

Alternatively, you may have an idea for a new package to cover an area not covered by existing packages. You can also reach out to discuss with any of the members on Slack or our GitHub project space.

Brief overview of what makes a tidyomics package

The goal of the tidyomics packages is not to re-implement tidyverse or Bioconductor functionality, nor to create new classes to replace Bioconductor classes. All of the tidyomics packages leverage existing functions and classes, and provide an interface (an API) between tidy verbs and Bioconductor objects.

An example is tidySummarizedExperiment, which does not re-implement or change the class of SummarizedExperiment objects, but instead provides a tibble abstraction of the object, which can be manipulated as if the SummarizedExperiment were in a tidy format. The underlying SummarizedExperiment maintains its class (unless there is a summarization operation for example) and can be operated on with existing Bioconductor methods as usual.

Tidyomics in scripts vs in package development

We are planning to develop more tutorials and instructional material on deciding between using tidyomics in scripts vs using tidyomics throughout package code. One consideration is compute speed, where certain matrix operations have highly optimized functions, e.g. the operations available in matrixStats. For repetitive tasks within the source code of a Bioconductor package, it may make more sense to use rowMeans or the like, rather than, say, group_by(row) |> summarize(mean = mean(counts)).

Typically in scripts, readability and extensibility is much more important than the speed of individual operations. In addition, for most analyses, the speed of tidyomics is equivalent to base R/Bioconductor (see tidyomics paper for speed comparisons).

Note that a tidyomics package need not make use of pipes or non-standard evaluation in its own source code, while it certainly would make use of these paradigms in its vignettes and examples. Feel free to post to the #tidiness_in_bioc channel on the Bioconductor community Slack if you want more guidance on this point.