NEWS.md

In NathanSkene/EWCE: Expression Weighted Celltype Enrichment

EWCE 1.11.5

Bug fixes

• check_bootstrap_args:
• annotation level for checking cell type names was previous hardcoded to level 1, now updated to match user input for annotation level.

EWCE 1.11.4

Bug fixes

• drop_uninformative_genes:
• Will now catch cases where expression matrix is a dataframe and convert to a matrix
• This was causing weird errors, see issue 92: https://github.com/NathanSkene/EWCE/issues/92
• Fix made in check_sce() function.

EWCE 1.11.2

New features

• bootstrap_enrichment_test:
• New args: standardise_sct_data=, standardise_hits=: let users have more control over data standardisation steps.
• check_ewce_genelist_inputs: updated accordingly.
• New arg: store_gene_data to avoid hitting memory limits.
• Modify test-bootstrap_enrichment_test_2.R to use test new args.

EWCE 1.11.1

Bug fixes

• ewce_plot() - Dendrogram not reordering cell types in plot
  • see issue
  • Occurs when ctd does not contain the plotting info
  • Fixed now and unit test added.
  • Note that cell type order on the x-axis is based on hierarchical clustering for both plots if make_dendro = TRUE for ewce_plot().

EWCE 1.9.3

Bug fixes

• generate_bootstrap_plots
  • Use stored gene_data object whenever possible.
  • Only show filtered celltypes.

EWCE 1.9.2

Bug fixes

• generate_bootstrap_plots
  • Missing available parameters for check_ewce_genelist_inputs() call.

EWCE 1.9.1

Bug fixes

• drop_uninformative_genes
  • Hash out DGE options. Somehow these got re-exposed to users in Bioc>=3.16.

EWCE 1.7.4

New features

• check_species:
  • Added new arg sctSpecies_origin_default
• standardise_ctd:
  • Can now specify sctSpecies_origin, which will be added to the metadata.

Bug fixes

• standardise_ctd:
  • When "annot" slot is not provided in the original CTD, a new one will be created from matrix columns names instead of assigning NULL.

EWCE 1.7.3

New features

• check_ewce_genelist_inputs/bootstrap_enrichment_test
  • New arg: sctSpecies_origin lets users clarify that their data originally came from mouse even when it is currently formatted as human orthologs. This is necessary for creating the appropriate background gene lists.
• Remove grDevices as dep entirely.
• fix_celltype_names
  • Add new arg make_unique to make this function easily usable for vectors where the same celltype appears multiple times.
• bootstrap_enrichment_test
  • Return gene-level scores based on adaptation of code from generate_bootstrap_plots. now stored as a list element named gene_data in data.table format.
• generate_bootstrap_plots
  • Revamp wrap code into reusable subfunctions.
  • Avoid resampling random genes when the gene data is stored in the bootstrap results as gene_data. It will also tell you which of these options it's using.
  • Save with ggsave instead if grDevices.
  • Facet by celltype instead of generating tons of separate plots.
  • Let users decide cutoff threshold with new arg adj_pval_thresh
  • Now returns a named list with the plots themselves ("plot") and the paths to where they're saved ("paths") rather than just a higher-level directory path in which users had to search for the right files (and didn't ever have access to the ggplot2 objects).
  • Show significance with barplot fill/color instead of asterices. Much easier to see now.
  • Change savePath arg to the more accurate save_dir. Expose appending BootstrapPlots to the user within the argument.
• generate_bootstrap_plots_for_transcriptome
  • Change savePath arg to the more accurate save_dir. Expose appending BootstrapPlots to the user within the argument.
  • Save with ggsave instead if grDevices.
• Standardise hits + hitGenes arg all to hits.
• Update hex:
  • Off load large source image from DALLE to Releases instead of including it within the package.

Bug fixes

• drop_uninformative_genes / generate_celltype_data
  • Pass verbose arg to matrix formatting functions.
• generate_controlled_bootstrap_geneset
  • Removed combinedGenes arg as it was not being used anywhere within.
• check_args_for_bootstrap_plot_generation
  • Removed unused args: ttSpecies, sctSpecies
• test-bootstrap_enrichment_test_2.R
  • "monkey_ctd" tests seems to be running more smoothly than before (not just getting NAs). This might have to with orthogene databases improving.
  • Reassuringly, "godzilla" tests still fail as expected :)
• Add tess/testthat/Rplots.pdf to .gitignore.

EWCE 1.7.1

New features

• Use rworkflows GHA.
  • Add rworkflows::use_badges to README.Rmd.
  • Remove Dockerfile (no longer necessary).
  • Make all 3 platforms (Linux, Mac, Windows) use Bioc dev, as ewceData (>=1.7.1) is now required, due to a fix made only in the development version of rtracklayer.
• Remove cowplot dependency.
• Replace all %>% with |>

Bug fixes

• calc_quantiles:
  • This function was only used in filter_variance_quantiles
  • Compare stats::ecdf vs. dplyr::ntile methods.
  • Remove from EWCE as it's not longer used anywhere.
• bin_columns_into_quantiles:
  • Rename arg matrixIn --> vec to reflect what the function actually does.
• filter_variance_quantiles:
  • Change to use bin_columns_into_quantiles instead of calc_quantiles to be consistent with how quantiles are handled in the rest of EWCE.
  • Updated tests in test-get_celltype_table.r to reflect that the number of genes filtered is unaffected by the normalization procedure (when quantiles are computed with stats::quantile).
• ewce_plot:
  • Celltypes were producing NAs because the names in the results were not always standardized in the same way as the CTD. Now this is done internally.
  • Celltypes were not ordered factors, meaning the dendrogram didn't line up correctly.
  • Switched from gridArrange/cowplot to patchwork.
  • Added dedicated unit tests file: test-ewce_plot.r

EWCE 1.7.1

New features

• Offline runs enabled with functions using reference datasets (from ewceData). These functions have the parameter localhub added to control this.

EWCE 1.5.8

Bug fixes

• GHA fix.

EWCE 1.5.7

Bug fixes

• orthogene dependency has been replacing user entered background gene list with one generated from all known genes when species across gene lists and reference dataset are the same. This has now been fixed.

EWCE 1.5.5

Bug fixes

• standardise_ctd:
  • Always force "specificity_quantiles" to be one of the matrices in each level.

EWCE 1.5.4

New features

• filter_ctd_genes
  • Now exported.
  • Can handle standardized CTD format.
• get_ctd_matrix_names: New function to get a list of all data matrices in CTD.

Bug fixes

• check_ewce_genelist_inputs:
  • User reported potential bug in code:https://github.com/NathanSkene/EWCE/issues/71
  • Fixed by removing conditional and instead always filtering out genes not present in CTD/SCT.
• standardise_ctd:
  • Add check_species()
  • Ensure all matrices become sparse when as_sparse=TRUE.
  • Generalize to matrices of any name.
• fix_celltype_names:
  • Ensure all celltype names are unique after standardization.

EWCE 1.5.3

New features

• genelistSpecies now passed to prepare_genesize_control_network in bootstrap_enrichment_test meaning gene list species will be inferred from user input.

EWCE 1.5.2

New features

• drop_uninformative_genes:
  • Expose new args: dge_method, dge_test, min_variance_decile
• merged_ctd: Actually merge the CTDs into one when as_SCE=FALSE.

Bug fixes

• Remove hard-coded file path separators (e.g. sprintf("%s/MRK_List2.rpt", tempdir())) to be more compatible with Windows.

EWCE 1.5.1

New features

• Made substantial updates to orthogene, so going through and making sure everything still works / is able to take advantage of new features (e.g. separation of non121_strategy and agg_func args, many:many mapping):
  • filter_nonorthologs: Pass up args from orthogene::convert_orthologs.
  • generate_celltype_data: @inheritDotParams
• Update GHA.
• Bump to R (>= 4.2) now that we're developing on Bioc 3.16.

Bug fixes

• Avoid downloading large "MRK_List2.rpt" file any more than is necessary for testing.

EWCE 1.3.3

New features

• method argument from orthogene::create_background and orthogene::convert_orthologs is now passed up as an argument to EWCE functions to give users more control. "homologene" chosen as default for all functions. "homologene" has fewer species than "orthogene" but doesnt need to import data from the web. It also has more 1:1 mouse:human orthologs.
• Include notes on mismatches between GitHub documentation and current Bioc release version.
• Allow bin_specificity_into_quantiles to set specificity matrix name produced.
• Merge GHA workflow yamls into one.

Bug fixes

• Add try({}) and error=TRUE to avoid "polygon edge not found" error in vignettes.

EWCE 1.3.1

New features

• Major changes: Pull Request from bschilder_dev branch.
• All functions can now use lists and CellTypeDatasets (CTD) from any species and convert them to a common species (human by default) via orthogene.
• Automated CTD standardisation via standardise_ctd.
• Can handle (sparse) matrices.
• Can create CTD from very large datasets using DelayedArray object class.
• All functions automatically create appropriate gene backgrounds given species.
• More modular, simplified vignettes.
• Additional gene pre-filtering options (DESeq2, MAST, variance quantiles).
• New/improved plotting functions (e.g. plot_ctd).
• Added example bootstrapping enrichment results as extdata to speed up examples (documented in data.R). Accessed via EWCE::example_bootstrap_results().
• Replaced GHA workflow with check-bioc to automatically: run R-CMD checks, run BiocCheck, and rebuild/deploy pkgdown site.
• Parallelised functions:
  • drop_uninformative_genes
  • generate_celltype-data
  • bootstrap_enrichment_test
• Added tests (multiple functions tests per file to reduce number of times ewceData files have to be downloaded):
  • test-DelayedArray
  • test-merge_sce
  • test-get_celltype_table
  • test-list_species
  • test-run_DGE
  • test-check_percent_hits
• Added function is_32bit() to all tests to ensure they don't get run twice on Windows OS.
• Added GitHub Actions workflows:
  • check-bioc-docker.yml: Runs CRAN/Bioc checks, rebuilds and pushes pkgdown website, runs and uploads test coverage report,
  • dockerhub.yml: Builds Bioconductor Docker container with EWCE installed, runs CRAN checks and (if checks are successful) pushes container to neurogenomicslab DockerHub.
• Removed docs folder, as the documentation website comes from the gh-pages branch now, and is automatically built by GHA workflow after each push to main branch.
• Added new exported function fix_celltype_names to help with standardising celltype names in alignment with standardise_ctd.
• generate_bootstrap_plots_for_transcriptome: Now supports any species (not just mouse or human).
  • Converts CTD and DGE table (tt) into output_species gene symbols.
  • Automatically generates appropriate gene background.
  • Faster due to now having the option to only generate certain plot types.
• Provide precomputed results from ewce_expression_data via new example_transcriptome_results function.
• Reduced build runtime and oversized vignettes by not evaluating certain code chunks.
  • Prevent extended vignette from running entirely.
• @return documentation for internal functions.
• Added more installation checks to GHA.
• Fixed inconsistent naming of unit test files: test_ ==> test-
• Removed DGE args in drop_uninformative_genes for now until we run benchmarking to see how each affects the EWCE results.
• Make bootstrap_plots function internal.
• Add report on how orthogene improve within- and across-species gene mappings in extended vignette.
• Record extra info in standardise_ctd output:
  • "species": both input_species and output_species
  • "versions": of EWCE, orthogene, and homologene

EWCE 1.0.0

New features

• EWCE v1.0 on Bioconductor replaces the defunct EWCE v1.3.0 available on Bioconductor v3.5.
• EWCE has been rendered scalable to the analysis of large datasets
• drop_uninformative_genes() has been expanded to allow the utilisation of differential expression approaches
• EWCE can now handle SingleCellExperiment (SCE) objects or other Ranged SummarizedExperiment (SE) data types and as input as well as the original format, described as a single cell transcriptome (SCT) object.

Deprecated & Defunct

• The following functions have been renamed to use underscore in compliance with Bioconductor nomenclature:
  • check.ewce.genelist.inputs
  • cell.list.dist
  • bootstrap.enrichment.test
  • bin.specificity.into.quantiles
  • bin.columns.into.quantiles
  • add.res.to.merging.list
  • prepare.genesize.control.network
  • prep.dendro
  • get.celltype.table
  • calculate.specificity.for.level
  • calculate.meanexp.for.level
  • generate.celltype.data
  • generate.bootstrap.plots
  • generate.bootstrap.plots.for.transcriptome
  • fix.bad.mgi.symbols
  • fix.bad.hgnc.symbols
  • filter.genes.without.1to1.homolog
  • ewce.plot
  • cells.in.ctd
  • drop.uninformative.genes

NathanSkene/EWCE documentation built on April 10, 2024, 1:02 a.m.