Submission for BioC 3.7

[nhejazi]

The package is now mostly ready to begin the process for submission to the Bioconductor project. The purpose of this issue is to keep track of changes that need to be made as we go through that submission process. For reference, documentation on the submission process is available here.

• [x] Open an issue at the BioC contributions GitHub repo. Note that only the package owner (@wilsoncai1992) can start this process.
• [x] Create push hook for BioC contributions issue to track updates to branch master on this repo.
• [x] Update CRAN version of origami to 0.8.2+
• [x] Flesh out class definition for adaptmle and wrapper function bioadaptest
• [x] Complete vignette with examples from Bioconductor data sets

[wilsoncai1992]

The submission issue is made

https://github.com/Bioconductor/Contributions/issues/628

[wilsoncai1992]

Initial Bioconductor 3.7 Review:

DESCRIPTION

• [x] Use the release version of roxygen2 (Wilson)
  • addressed in #41

NAMESPACE

• [x] You may want to separate the "shinyprint" functionality into a different package? (Wilson)
  • addressed in #41

R

• [x] Avoid using cryptic argument names, use names that are descriptive and helpful for the end user (Wilson)
  • addressed in #41
• [ ] The data_adapt class is based on a list which is not the best structure to use as you can't really enforce any validity checks to the components of the class as easily as you would with an S4 class. (Nima)
  • proposal: to be addressed in a future release -- pending the need to experiment with exactly how comfortable non-Bioconductor users are with S4 or R6 classes rather than ad-hoc S3 classes.
• [x] Avoid the use of 1:n constructions (Wilson)
  • addressed in #41
• [x] Use the allocate and fill method, e.g., ls <- vector("list", ncol(adaptY_composition)) (Nima)
  • addressed in #43
• [x] Avoid using grep to find colnames in cv_param_est. Let the user provide the colname directly from the argument. (Wilson)
  • addressed in #41
• [x] Replace sapply calls with the more robust vapply (Nima)
  • addressed in #43
• [x] You shouldn't need to "wrangle" the supported SummarizedExperiment class for your main function, it should just work with it as part of the integration process. In other words, it should make use of the SummarizedExperiment API rather than the assay(x); t(x) and rownames(x) <- colnames(x) <- NULL version. (Nima)
  • addressed by having the adaptest function wrangle RangedSummarizedExperiment objects differently from data.frame or Matrix, making the analysis function more extensible (#43)
• [x] A proper API should not use @ accessor functions. See: http://bioconductor.org/developers/how-to/efficient-code/ (Nima)
  • addressed by addition of get_results_adaptmle accessor function in #43

data

• [x] Provide descriptive names for your data, Y and A can mean anything. (Wilson)
  • addressed in #41

vignette

• [x] It looks like you commented BiocStyle out. (Wilson)
  • addressed in #41

• [x] Please add an installation chunk. It doesn't have to run. (Wilson)

  • addressed in #41

  • NOTE (for reviewer): currently, we use the standard S3-style classed list as the core structure for across-community compatibility of the statistical routine exposed by this software. It is our aim to make this new statistical technique easy to use for the Bioconductor community via the bioadaptest wrapper function while still making the core software easy to inspect for those well-versed in the methodology produced by our statistical community but yet unfamiliar with Bioconductor and related best practices (though we do hope to slowly remove this part of the code base and replace with S4, or maybe R6, as the package picks up users).

[nhejazi]

Closing since the package has been accepted for BioC 3.7