Skip to content

Latest commit

 

History

History
161 lines (119 loc) · 10.1 KB

NEWS.md

File metadata and controls

161 lines (119 loc) · 10.1 KB

jackstraw 1.3.2 (2020-12-23)

  • Fixed lots of minor things that caused check errors
    • Vignette: minor edits to fix recent build errors
    • Moved external package imports from R code to DESCRIPTION, adding missing packages, and fixed namespaces.
    • Corrected function arguments and usage discrepancies.
  • Added this NEWS.md file to track changes to the package.

jackstraw 1.3.3 (2021-03-09)

Overview:

  • Added unit tests to almost all functions
    • Basic uses are covered (default optional parameters), and some but not all advanced uses
  • Minor user-facing changes:
    • Moved mandatory arguments to the front (no backward-incompatible problems since these arguments had to be called by name unless the earlier optional arguments were also present and were passed unnamed (rare))
    • Added more informative error messages for missing mandatory arguments, objects with wrong classes, and covariate dimension disagreements
    • Minor bug fixes (some s = 1 edge cases)
    • Eliminated all non-error/warning verbosity when verbose = FALSE.

Additional details:

  • Function jackstraw_cluster
    • Added option pool (default TRUE) to calculate p-values by pooling null statistics (to match jackstraw_kmeans option and default). Previously the hardcoded behavior matched pool = FALSE.
  • Functions jackstraw_lfa, jackstraw_alstructure, jackstraw_pca, jackstraw_rpca, jackstraw_irlba, jackstraw_subspace:
    • Argument r is now second argument.
      • lfa, alstructure, subspace versions: r does not have a default value and it is mandatory.
  • Functions jackstraw_kmeans, jackstraw_kmeanspp, jackstraw_MiniBatchKmeans, jackstraw_pam, jackstraw_cluster:
    • Debugged s = 1 edge case: null data used to be centered incorrectly. Bug only occurred in combination with center = TRUE (default).
  • Internal (unexported) function RSS now returns actual residual sum of squares. This change does not affect any exported functions that use it. Previously RSS calculated a normalized version (equal to 1 - R^2), but this normalization canceled out in FSTAT (its only downstream use), so the normalization had no user-facing effect.

Exclusive list of functions without unit tests (all are redundant with other packages, so they are candidates for removal in the near future):

  • Exported:
    • lfa.corpcor (redundant with lfa::lfa)
    • pi0est_bootstrap (redundant with qvalue::pi0est)
  • Internal, no exported backward dependencies:
    • getp (redundant with qvalue::empPvals)
    • devdiff_parallel (redundant with gcatest::gcat.stat)

jackstraw 1.3.4.9000 (2021-04-14)

  • Functions jackstraw_pca, jackstraw_rpca, jackstraw_irlba: Corrected documentation (parameter r1 was incorrectly described as PC in parts of the documentation. Thanks to Djordje Bajić (GitHub username djbajic) for reporting this error!
  • Removed option seed from all functions that had it. For the same behavior, call set.seed(seed) before calling the function.
  • Functions jackstraw_lfa and jackstraw_alstructure: removed devR option.
  • Removed redundant functions
    • lfa.corpcor: same as lfa::lfa with option override = TRUE
    • pi0est_bootstrap: redundant with qvalue::pi0est with option pi0.method = 'bootstrap'
    • dev.R (internal; functionality implemented in package gcatest)
    • devdiff_parallel (internal; redundant with gcatest::gcat.stat)
    • getp (internal; redundant with qvalue::empPvals)

jackstraw 1.3.5.9000 (2021-05-14)

  • Function jackstraw_lfa now accepts genotypes input as BEDMatrix objects. In this case, the function operates on a low-memory mode, keeping data on disk rather than memory as much as possible, and writes permuted data into temporary files as well. To enable this mode, the BEDMatrix and genio packages are now dependencies. Note only jackstraw_lfa supports BEDMatrix because lfa supports it too (most recent fork; see below).
  • Removed function devdiff, which is redundant (and replaced internally) with gcatest::delta_deviance_lf, a function that supports more special cases, including genotypes accessed through a BEDMatrix object. The only internal dependencies were jackstraw_lfa and jackstraw_alstructure.
  • Updated README.md to instruct users to install the most updated forks of lfa and gcatest on GitHub (under username alexviiia), rather than the Bioconductor versions that are lacking critical updates.

jackstraw 1.3.5.9000 (2021-06-21)

  • Version was not bumped accidentally
  • Removed old vignette
  • Fixed a small error in jackstraw_lfa example

jackstraw 1.3.6.9000 (2022-02-08)

  • All jackstraw_* functions now return NA p-values for NA statistics.
    • Before NA statistics resulted in p-values of 1 instead, which is what qvalue::empPvals returns. Now an internal wrapper function ensures the desired behavior.
  • Removed two jackstraw_pam toy example unit tests that failed often due to colinearity.
  • Reformatted this NEWS.md slightly to improve its automatic parsing.

jackstraw 1.3.9.9000 (2022-11-28)

  • Heavy updates to README.md
  • Updated summary in DESCRIPTION
  • Updated paper citations and URLs across function documentation
  • Bumped version to exceed CRAN versions (which are currently on a separate branch due to temporary issues with dependencies lfa and gcatest)

jackstraw 1.3.10.9000 (2023-08-03)

  • Function jackstraw_MiniBatchKmeans: internally added suppressWarnings wrapper around ClusterR::predict_MBatchKMeans to silence a misleading deprecation warning. The function as a whole is not deprecated, but its output for a case we don't use (fuzzy=TRUE) is changing in the future. Warning message now being suppressed:
    • predict_MBatchKMeans() was deprecated in ClusterR 1.3.0.
    • Beginning from version 1.4.0, if the fuzzy parameter is TRUE the function predict_MBatchKMeans will return only the probabilities, whereas currently it also returns the hard clusters
  • Removed strict dependence of the alstructure R package (on GitHub only), user now has to provide the function to apply to data for jackstraw_alstructure to work.
  • Documentation
    • Corrected citation links for two similar papers
    • Function documentation (.Rd files) was actually updated as described in the last version (last commit had source .R files edited but not their .Rd counterparts).
    • Added a recent missing NEWS.md entry
    • Replaced two links that gave errors in README.md

Changes manually ported from latest CRAN branch:

  • Minor non-code edits
    • Spell checked documentation
    • Changed to single maintainer (Neo)
    • Removed VignetteBuilder: knitr since there's no vignette anymore
    • Fixed broken or outdated URLs
    • Removed package dependency parallel which is not being used.
    • Removed many package suggested dependencies (knitr, rmarkdown, ggplot2, mutoss, Matrix, gridExtra, cowplot, scales, formatR) that were only used in a vignette currently not being built.
    • Reduced example dimensions for jackstraw_irlba and jackstraw_rpca by 5 to keep their runtime low.

jackstraw 1.3.11.9000 (2023-08-03)

  • Function jackstraw_lfa added option permute_alleles, to change default genotype permutation to an allele-level permutation algorithm that results in much more Binomial data, particularly when input data is highly structured.

jackstraw 1.3.12.9000 (2024-08-01)

  • README edited installation instructions to reflect that dependencies lfa and gcatest are now updated on Bioconductor (no need to install GitHub versions) and alstructure is now optional.
  • Function jackstraw_lfa changed option permute_alleles to default to TRUE, which performs much better, and documentation now recommends against changing that option except for research purposes.
  • Corrected typos throughout, and marked code as such to prevent false positives.

jackstraw 1.3.13 (2024-08-01)

  • CRAN submission
  • DESCRIPTION replaced Author and Maintainer with Authors@R field
  • Function jackstraw_lfa silenced warnings that occurs only if input is BEDMatrix object and OS is Windows, in which case some temporary files cannot be deleted during the R session.
  • Created cran-comments.md

jackstraw 1.3.14 (2024-08-29)

  • README edited. Use cases as a separate subsection.
  • Description in DESCRIPTION edited.

jackstraw 1.3.15 (2024-08-30?)

  • Use DOIs instead of publication URLs, as required by CRAN.

jackstraw 1.3.16 (2024-09-12)

  • Moved Bioconductor dependencies (qvalue, lfa, and gcatest) from Imports to Suggests, as required by CRAN. Dependent functions will now only work if those packages are installed:
    • Package qvalue is required for function pip.
    • Package lfa is required for functions pseudo_Rsq and efron_Rsq, and in practice for jackstraw_lfa too.
    • Package gcatest is required for functions jackstraw_alstructure and jackstraw_lfa.
    • Function jackstraw_lfa now requires option FUN to be defined and provided by the user (it used to default to the lfa function)
    • Internal function empPvals is now a copy of the function of the same name in the qvalue package, with minor adaptations. This is to ensure most jackstraw functions, which depend on empPvals, work without qvalue.

jackstraw 1.3.17 (2024-09-16)

  • m increased in unit tests to pass building in CRAN r-devel-windows-x86_64
  • README edited

jackstraw 1.3.17.9000 (2024-11-08)

  • Added two functions for calculating high precision p-values from the data output by jackstraw_lfa and jackstraw_alstructure:
    • ncp_est: estimates parameter of non-central chi squared distribution from samples.
    • pvals_nc_chisq: uses ncp_est to fit a non-central chi squared model to the Jackstraw null statistics, then calculates p-values from this model for the observed statistics.
  • New package dependency stats4 added to calculate maximum likelihood estimates of the non-centrality parameter.
  • Internally, documentation now uses Markdown shortcuts.

jackstraw 1.3.18.9000 (2024-12-12)

  • Minor documentation updates to functions jackstraw_lfa and pvals_nc_chisq

jackstraw 1.3.19.9000 (2024-12-19)

  • P-values of 0 from a non-central chi squared model are set to .Machine$double.xmin in pvals_nc_chisq