Vignettes quality
Closed this issue · 6 comments
Still poor in terms of description quality and general logic, language need some polishing too. Do link me here once you have improced those, before closing this issue.
@caravagn here you are, glad if you can check the new version of vignettes.
devtools::document()says that examples are missing;pkgdown::build_site()reports YAML warnings;- there were things repeated across vignettes (opening parts) that did not make sense, I made a special vignette [TAPACLOTH.Rmd]
- I don't understand the columns
purityandsampleare repeated - S3 methods can be more informative and show nice outputs etc while they do not now.
I don't think this current status is up to our standards yet.
I am pushing to fix these.
All the things are, from a software engineering, really illogical.
Example: fit is sometimes a tibble, other times a Bmix object, there is no consistency and well-behaved practices. There have to be uniform SW-practices, otherwise maintaining code is a horrible pain. For instance, how can I run on the same test both the purity estimation and the clonality test?
The current version is close to a spaghetti code.
OK thanks! I made it such that the output of both run_classifier and estimate purity is an S3 object of class TAPACLOTH, that always contains the input data in data, possibly with new columns (e.g. class_binom, class_terzile or purity_bmix), and either a classifier or a purity_estimate object or both. These two contain for example the input paramaters and/or plots and fits.
Everything should be consistent.
If so, I close the issue.