dyngen 1.0.0

[rcannood]

This version mostly upgrades dyngen's ease-of-use, such as better vignettes, conversion functions for working with dyngen datasets in other packages, and more useful ways of specifying platform-specific parameters (i.e. number of cores and cache location). Perhaps more excitingly, the dyngen documentation is more readable online at https://dyngen.dynverse.org!

BREAKING CHANGES

• wrap_dataset(): Now returns a list instead of a dyno object. Use as_dyno(model) or wrap_dataset(model, format = "dyno") to replicate previous behaviour.

NEW FEATURES

• Added functions for converting the dyngen output to various data formats: as_anndata() for anndata, as_sce() for SingleCellExperiment, as_seurat() for Seurat, as_dyno() for dyno, as_list() for a simple list object.

• wrap_dataset(): Added 'format' argument which allows choosing the output format (dynwrap should be an Imports dependency, not Suggests #28).

• The default number of cores used can be set by adding options(Ncpus = ...) to your Rprofile.

• The default cache folder for dyngen can be set by adding options(dyngen_download_cache_dir = ...) to your Rprofile.

• Combine similar models with different outputs using the combine_models() function.

• Store the timings throughout the dyngen execution. Extract the timings from a model using get_timings().

MAJOR CHANGES

• generate_experiment(): Map count density of reference dataset to simulation expression before sampling molecules.
  Parameters are available for toggling off or on the mapping of the reference library size & CPM distribution.

MINOR CHANGES

• initialise_model(): Change defaults of num_cores and download_cache_dir
  to getOption("Ncpus") and getOption("dyngen_download_cache_dir") respectively.

• generate_experiment(): Drastically speed up sampling of molecules.

BUG FIX

• as_dyno(): Fix drop = FALSE bug when only one cell is being sampled.

• Removed names from feature ids in feature info (unname(model$feature_info$feature_id)). Thanks @milanmlft!

DOCUMENTATION

• Added and extended vignettes:

  • Advanced: Simulating batch effects
  • Advanced: Simulating a knockout experiment
  • Advanced: Running dyngen from a docker container
  • Advanced: Constructing a custom backbone
  • Advanced: Tweaking parameters
  • Advanced: Comparison of characteristic features between dyngen and reference datasets

• Created a website at https://dyngen.dynverse.org using pkgdown.

• Shortened examples to reduce r cmd check time.

[codecov]

Codecov Report

Merging #27 (4f77ead) into master (0967ed8) will increase coverage by 5.46%.
The diff coverage is 81.38%.

@@            Coverage Diff             @@
##           master      #27      +/-   ##
==========================================
+ Coverage   63.49%   68.95%   +5.46%     
==========================================
  Files          20       22       +2     
  Lines        2257     2654     +397     
==========================================
+ Hits         1433     1830     +397     
  Misses        824      824              

Impacted Files	Coverage Δ
R/0b_modnet_pieces.R	0.00% <ø> (ø)
R/0c_backbone_lego.R	41.52% <ø> (ø)
R/0d_backbones.R	1.31% <0.00%> (ø)
R/rnorm_bounded.R	100.00% <ø> (ø)
R/simtime_from_backbone.R	100.00% <ø> (ø)
R/timings.R	25.00% <25.00%> (ø)
R/plotting.R	81.14% <63.49%> (-4.25%)	⬇️
R/1_initialisation.R	90.00% <73.33%> (-10.00%)	⬇️
R/generate_dataset.R	79.24% <80.00%> (-3.37%)	⬇️
R/7_experiment.R	76.14% <80.64%> (ø)
... and 15 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 0967ed8...4f77ead. Read the comment docs.