NEWS.md

In neurogenomics/EpiCompare: Comparison, Benchmarking & QC of Epigenomic Datasets

CHANGES IN VERSION 1.3.4

New features

• Example report:
• Delete report/ folder and upload to Releases instead: https://github.com/neurogenomics/EpiCompare/releases
• Add Rscript to replicate example report in inst/examples.
• EpiCompare
• New arg add_download_button.
• Always keep download button for post-processed peak files.
• `save_output``
• Change .txt suffixes to more informative .csv suffixes for saved files.
• Create EpiArchives and offload report to there:
• Updated vignettes/example_report.Rmd so that it just renders the markdown landing page for EpiArchives.

Bug fixes

• README.Rmd:
• Fix broken link to example report.
• test-EpiCompare.R:
• Fix issue with PNG save path.
• Make separate subfolders for each set of tests.
• test-output_files.R
• Make separate subfolders for each set of tests.
• plot_enrichment
• Conditionally generate plots only when enrichment results aren't NULL.
• Return KEGG/GO enrichment results as well as the plots.
• Add workers <- check_workers(workers = workers) to all functions that take workers to handle workers=NULL properly.

CHANGES IN VERSION 1.3.3

New features

• download_button:
• Saves and downloads files.
• prepare_blacklist:
• Auto-selects appropriate blacklist, or returns user-specified option.
• EpiCompare(blacklist=NULL) is now the default.
• prepare_genome_builds:
• Update to handle supplying builds for "peakfiles" and "reference" but not "blacklist" (so long as the blacklist arg is not a user-supplied GRanges object)
• Added mm9_blacklist
• Made more plots interactive:
• width_boxplot
• plot_enrichment
• plot_ChIPseeker_annotation
• overlap_stat_plot
• Name elements in output list.
• Change annotation arg to more informative txdb arg, and set default to NULL, which ChIPseeker functions will automatically handle.
• New function as_interactive:
• Help standardise this.
• New EpiCompare::EpiCompare arguments:
• error: keep knitting even on errors.
• tss_distance: upstream/downstream of TSS.
• quiet: knit quietly
• Rename 'test-EpiCompare_combinations.R' --> 'test-EpiCompare.R'
• Separate test-generalMetrics_functions.R into function-specific test files.
• Separate test-peakOverlap_functions.R into function-specific test files.
• Make fancy header with new func:
• report_header()
• Create EpiCompare command code as text:
• report_command()
• width_boxplot:
• Make more efficient with data.table and lapply
• Update hex sticker to match custom.css palette.
• README.Rmd
• Collapse more detailed sections.

Bug fixes

• tss_plot:
• Fix examples/tests after Sera updated the arguments.
• Pass upstream/downstream to ChIPseeker::getTagMatrix
• Make interactive
• Name plots in list
• Remove unnecessary extra level of list nesting.
• Make documentation width <80 lines where possible.
• EpiCompare.Rmd
• Remove methods::show from all parts
• Name all chunks
• Make explanations more clear
• Add table of contents for main 3 sections.
• Fix header levels
• Set results='asis' globally instead of in each chunk header.
• Automatically number sections with yaml arg: number_sections: true
• Omit specific headers from numbering system with {-} tags.
• Add custom.css
• plot_chromHMM:
• Error in (function (classes, fdef, mtable) unable to find an inherited method for function ‘annotateWithFeatures’ for signature ‘"SimpleGRangesList", "list"’
• Misleading error message; was actually due to chromHMM_annotation not being converted from a list to a GRangesList.
• Change yaml arg peakfile --> peakfiles to be consistent with other variables.

CHANGES IN VERSION 1.3.1

New features

• Replace badger with rworkfows:
  • Use rworkflows::use_badges
• New helper functions:
  • precision_recall_matrix
  • report_time
• overlap_upset_plot:
• Switched out UpSetR for ComplexUpsetto show percentages.
• Moved up dep checks to beginning of function.
• Handle bug with heatmaply by checking args where it might be used:
• check_heatmap_args
• tss_plot:
• Add unit tests
• Drastically reduce example/test runtime by setting upstream=50
• compute_corr:
• Reduce example runtime by setting bin_size = 200000 (takes <2s).

Bug fixes

• Fix typo in EpiCompare docs: "hg38 blacklist dataset"
• Avoid explicitly specifying "/" in paths to help cross-platform testing.
• tss_plot:
• Use parallel::detectCores-1 by default to set workers, but set to 1 in examples/tests to meet CRAN/Bioc standards.

CHANGES IN VERSION 1.1.4

New features

• Add back example report html:
  • Put it in the main dir
  • Add it to the .Rbuildignore: report/EpiCompare_example.html
  • Add a new vignette that renders the HTML from the pre-saved file.
• Remove Dockerfile, as it's no longer necessary with the updated version of rworkflows

Bug fixes

• Add @returns to group_files function.
• Add all authors to vignettes.

CHANGES IN VERSION 1.1.3

New features

• New function: predict_precision_recall
  • Added unit tests.
• compute_corr and precision_recall now save outputs, including when run via EpiCompare Rmarkdown script.
• Make subfunctions for plot_precision_recall:
  • plot_precision_recall_prcurve
  • plot_precision_recall_f1
• rebin_peaks:
  • Allow users to specify sep between genomic coordinates in rownames.

Bug fixes

• Update gather_files to match new Picard file scheme in nf-core/cutandrun 3.0.

CHANGES IN VERSION 1.1.2

New features

• rebin_peaks:
  • Added arg drop_empty_chr to automatically drop chroms that aren't in any of the peakfiles.
  • Added "score" as one of the default intensity_cols in all relevant functions.
  • Make examples use 5000bp bins to speed up.
• translate_genome:
  • Add default_genome arg to handle genome=NULL.
• bpplapply:
  • New exported function to automate handling of known issues with BiocParallel across OS platforms.
  • Enable users to specify their own apply function.
• get_bpparam: Add args to allow users to choose which BiocParallel func to use.
• checkCache: Make default arg cache=BiocFileCache::BiocFileCache(ask = FALSE) to skip user input during runtime.
• precision_recall:
  • Change increment_threshold arg to n_threshold arg, using the seq(length.out=) feature to avoid accidentally choosing an inappropriately large increment_threshold.
• gather_files:
  • Replace iterator with bpplapply.
  • Pass up args from bpplapply.
  • Provide warning message, not error, when 0 files found. Returns NULL.
  • Add "multiqc" as a search option.
  • Add dedicated subfunctions for reading in a variety of nf-core/cutandrun outputs files: read_picard,read_multiqc,read_bowtie, read_trimgalore,read_bam,read_peaks
  • Add file paths to each object.
  • Add new arg rbind_list.
• rebin_peaks/compute_corr: -Change defaultbin_size from 100 --> 5kb to improve efficiency and align with other defaults of other packages (e.g Signac).
• tss_plot:
  • Pass up more arg for specifying upstream/downstream.
• EpiCompare: Pass up new args:
  • bin_size
  • n_threshold
  • workers

Bug fixes

• Fix rebin_peaks unit tests.
• Fix pkg size issue by adding inst/report to .Rbuildignore.
• EpiCompare wasn't being run when reference was a single unlisted GRanges object because it was indeed length>1, but the names were all NULL. Now fixed.
• plot_precision_recall: Set default initial_threshold= to 0.
• Switch from BiocParallel to parallel, as the former is extremely buggy and inconsistent.

CHANGES IN VERSION 1.1.1

New features

• check_genome_build: Add translate_genome as prestep.
• rebin_peaks:
  1. Move all steps that could be done just once (e.g. creating the genome-wide tiles object) outside of the BiocParallel::bpmapply iterator.
  2. Ensure all outputs of BiocParallel::bpmapply are of the same length, within the exact same bins, so that we can return just the bare minimum data needed to create the matrix (1 numeric vector/sample).
  3. Instead of rbinding the results and then casting them back into a matrix (which is safer bc it can handle vectors of different lengths), simply cbind all vectors into one matrix directly and name the rows using the predefined genome-wide tiles.
  4. Because we are no longer rbinding a series of very long tables, this avoids the issue encountered here #103. This means this function is now much more scalable to many hundreds/thousands of samples (cells) even at very small bin sizes (e.g. 100bp).
  5. A new argument keep_chr allows users to specify whether they want to restrict which chromosomes are used during binning. By default, all chromosomes in the reference genome are used (keep_chr=NULL), but specifying a subset of chromosomes (e.g. paste0("chr",seq_len(12))) can drastically speed up compute time and reduce memory usage. It can also be useful for removing non-standard chromosomes (e.g. "chr21_gl383579_alt", "chrUns...", "chrRand...").
  6. As a bonus, rebin_peaks now reports the final binned matrix dimensions and a sparsity metric.
• compute_corr:
  • Added unit tests at different bin sizes.
  • Allow reference to be NULL.
• Updated README to reflect latest vesion of EpiCompare with gather_files.

Bug fixes

• Bumped version to align with Bioc devel (currently 1.1.0).
• compute_percentiles:
  • Making default initial_threshold=0, so as not to assume any particular threshold.
• rebin_peaks:
  • Addressed error that occurs when there's many samples/cells with small bins.
• plot_precision_recall: Don't plot the reference as part of the PR curve.

CHANGES IN VERSION 0.99.21

Bug fixes

• Changed terminology from "epigenetic" to "epigenomic"
• Updated README to include precision_recall_plot and corr_plot
• Removed bugs in html report
• Added example EpiCompare report in inst/report/EpiCompare_example.html

CHANGES IN VERSION 0.99.20

New features

• Upgraded liftover_grl and added genome standardization.
  • Enable cross-species liftover from/to mm10 and mm9 --> hg19 and hg38.
  • Subfunctionalize get_chain_file.
  • Add merge_all option.
• Support mm10/mm9 as output_build options.
• Removed dplyr.
• Moved plyranges to Suggests.
• plot_precision_recall:
  • New exported function to create precision-recall plots from MACS2, MACS3, HOMER, or SEACR peak files.
  • Added unit tests.
  • Added to EpiCompare html report.
  • Added EpiCompare(precision_recall_plot=) param and documented.
• Add ISSUE templates.
• Include code in html report (collapsed by default).
• Add correlation matrix/plot functionality.
• Add compute_consensus_peaks() as function for preprocessing peak files.
  • Add group_files() function to help assign each peakfile to a group based on substring searches.
• EpiCompare:
  • Return paths to HTML reports.
  • Automatically open report in browser or rstudio.
• Add Docker vignette and advertise in README.

Bug fixes

• Made BiocParallel functions compatible with Windows.
• Organize author fields in DESCRIPTION.
• Fix typos in README.
• Remove threshold=1 from list of thresholds to test in precision-recall curves.
• Set first chunk in EpiCompare.Rmd as echo=FALSE instead of include=FALSE so the output messages will still be printed (without showing the code).
• Remove here from Suggests.
• Fix directory creation in EpiCompare::EpiCompare.

CHANGES IN VERSION 0.99.19

Bug fixes

• Simplified loops with mapply/lapply.

New features

• EpiCompare: accepts multiple reference files - creates individual reports for each reference. Added timing feature.
• save_output(): this function saves all plots and tables generated by EpiCompare. Also saves interactive heatmaps. Used in EpiCompare.Rmd.
• fig_length(): This function outputs dynamic figure height/width depending on the number of items. Used in EpiCompare.Rmd.

CHANGES IN VERSION 0.99.18

Bug fixes

• prepare_reference: Validate reference input before passing to next step.
• Pass named list to genome_build to allow for different builds between reference and peaklist.
• Liftover blacklist to match GRanges list it's being used to filter in tidy_peakfile.
• Ensure all names are unique in peaklist and reference.
• gather_files:
  • Avoid gathering duplicates peak files from nf-core/cutandrun.
  • Add progress bar.
  • Add report at the end.
  • Add extra arg return_paths to return only the paths without actually reading in the files.

New features

• Overhaul how EpiCompare handles genome builds:
  • New argument genome_build_output allows users to specify which genome build to standardise all inputs to.
  • genome_build can now take a named list to specify different genome builds for peakfiles, reference, and blacklist.
  • Added functions to parse and validate all genome build-related arguments.
• Remove unnecessary deps.
• Use data.table to read/write tables.
• prepare_peaklist:
  • Simplified code.
  • Added arg remove_empty to automatically drop any empty elements.
  • Embed check_list_names within.
• plot_chromHMM:
  • Can return data as well with return_data.
  • Performs liftover on chromHMM data instead of the peaklist.

CHANGES IN VERSION 0.99.17

Bug fixes

• Make output_dir creation recursive and without warnings.
• Add new params to Code section of rmarkdown output.

New features

• Add new peaklist length check to prepare_peaklist.
• New check functions:
  • check_genomebuild: ensure necessary packages installed and that "genomebuild" is valid.
  • check_cell_lines
• liftover_grlist: Dedicated liftover function, exported.
• Document checkCache.
• get_chromHMM_annotation can now take a list of cell lines as an argument.

CHANGES IN VERSION 0.99.16

Bug fixes

• Fix GHA pkgdown building:
  • The newest version of git introduced bugs when building pkgdown sites from within Docker containers (e.g. via my Linux GHA workflow). Adjusting GHA to fix this.

CHANGES IN VERSION 0.99.3

New Features

• New functions with examples/unit tests:
  • import_narrowPeak: Import narrowPeak files, with automated header annotation using metadata from ENCODE.\
  • gather_files: Automatically peak/picard/bed files and read them in as a list of GRanges objects.\
  • write_example_peaks: Write example peak data to disk.
• Update .gitignore
• Update .Rbuildignore

CHANGES IN VERSION 0.99.1

New features

• New parameter in EpiCompare:
  • genome_build: Specify the genome build, either "hg19" or "hg38". This parameter is also included in plot_chromHMM, plot_ChIPseeker_annotation, tss_plot and plot_enrichment.

CHANGES IN VERSION 0.99.0

New Features

• EpiCompare submitted to Bioconductor.

neurogenomics/EpiCompare documentation built on April 30, 2024, 3:58 p.m.