Changes in version 2.2.3.9000                      

                 Changes in version 2.2.3 (2026-05-30)                  

Improvements

  - CRAN Compliance: Fixed policy violations regarding home directory
    usage and internet access during checks.
      - Default cache directory changed from tools::R_user_dir() to a
        session-specific temporary directory.
      - Added internet connectivity check for both Google and Baidu (for
        Chinese users).
      - Fixed example handling: replaced if(interactive()) with proper
        \dontrun{} blocks for genuinely non-runable examples
        (network-dependent, user files required, Suggested packages).
        CRAN clarified that if(interactive()) does NOT skip examples in
        checks.
      - Minimized \dontrun{} usage to 22 blocks with documented reasons;
        functions that can run offline use simulated data.
      - Users can still opt-in to persistent caching via
        set_iobr_cache_dir() or options(IOBR.cache_dir = ...).

Bug Fixes

  - Fixed count2tpm() to return NULL gracefully when annotation data is
    unavailable offline.

                 Changes in version 2.2.2 (2026-05-25)                  

Improvements

  - Supported non-GitHub mirrors and added fallback.

Bug Fixes

  - Fixed GitHub Actions CI failure on r-devel by correcting the
    http-user-agent configuration and removing redundant dependency
    specifications.
  - Added BiocManager to Suggests in DESCRIPTION to improve Bioconductor
    package resolution and address dependency availability issues
    reported in CRAN checks.

                 Changes in version 2.2.1 (2026-05-15)                  

New Features

  - Custom Cache Directory: Added support for customizing the download
    cache location via options(IOBR.cache_dir = "your/path"). New
    functions:
      - get_iobr_cache_dir() - Get current cache directory path
      - set_iobr_cache_dir(path) - Set custom cache directory
      - reset_iobr_cache_dir() - Reset to default system cache location
        This enables users to store cached data on shared network drives
        or any preferred location for easy access across sessions.

Improvements

  - Modified content of some files to adapt to portal display and
    improve minor errors and prompts (#135).

                 Changes in version 2.2.0 (2026-04-22)                  

CRAN Policy Fix (Resubmission)

  - Fixed all functions that wrote to the user's home filespace (working
    directory) by default. Writing now only occurs when an explicit path
    / save_path / output.dir is provided.
  - Affected functions: eset_distribution(), find_outlier_samples(),
    iobr_cor_plot(), sig_pheatmap(), sig_box_batch(), plotPurity(),
    IPS_calculation(), find_mutations(), sig_gsea(), get_cor(),
    batch_sig_surv_plot(), format_signatures(), and creat_folder().
  - Examples updated to use tempdir() when file writing is demonstrated.

Bug Fixes

  - find_mutations(): Fixed semantic naming error where file_name
    variable was used to store directory paths. Renamed to output_dir
    for clarity. Fixed ggsave() parameter order issues.
  - iobr_cor_plot(): Fixed ggsave() parameter order issues. The correct
    order is filename first, then plot.
  - surv_group(): Fixed ggsave() parameter order issues.
  - roc_time(): Fixed ggsave() parameter order issues.
  - batch_sig_surv_plot(): Changed default save_path from
    file.path(tempdir(), "Multiple-KM-plot") to NULL to prevent
    automatic directory creation.
  - format_signatures(): Changed parameter name from output_name to
    output_path for consistency. Added validation requiring output_path
    when save_signature = TRUE.
  - find_outlier_samples(): Added validation requiring project when save
    = TRUE.
  - plotPurity(): Changed default output.dir from
    "estimated_purity_plots" to NULL.

Major Changes

  - CRAN Size Compliance: Moved large datasets (>5MB total) from
    R/sysdata.rda and data/ to GitHub Releases to meet CRAN package size
    requirements. Data is now downloaded on-demand and cached locally.
  - New Data Management System:
      - Added load_data() function for unified data access (supports
        sysdata, exported data, and GitHub-hosted datasets)
      - Added download_iobr_data() with multiple mirror support (GitHub,
        ghproxy.vip, gh-proxy.org, ghfast.top)
      - Added add_iobr_mirror() for custom mirror configuration
      - Added list_github_datasets() to show available remote datasets
      - Added clear_iobr_cache() to manage downloaded data
  - Automatic Caching: Downloaded datasets are cached in user's cache
    directory and persist across package updates.
  - Fixed count2tpm() to handle symbol-based gene IDs more robustly.

Data Migration

The following datasets are now hosted on GitHub and downloaded on first
use:

  - Reference matrices: BRef, TRef, lm22
  - Annotations: anno_gc_vm32, anno_grch38, anno_hug133plus2,
    anno_illumina, anno_rnaseq
  - Example datasets: tcga_stad_sig, imvigor210_sig, eset_stad,
    sig_stad, eset_gse62254, etc.
  - Gene sets: hallmark, kegg, go_bp, go_cc, go_mf, reactome,
    msig_immune, msig_sc
  - Cell markers: cancer_type_genes, cellmarkers, common_genes,
    immuneCuratedData, ips_gene_set, SI_geneset, mRNA_cell_default,
    mus_human_gene_symbol, onco_sig, PurityDataAffy
  - Reference data: xCell.data, quantiseq_data
  - Signatures: signature_collection_citation, signature_metabolism,
    signature_sc, signature_tumor

Internal Updates

  - Updated all internal code to use load_data() instead of direct
    object references for migrated datasets
  - Updated globalVariables.R to reflect data migration
  - Enhanced error messages with manual download instructions when all
    mirrors fail

                        Changes in version 2.1.1                        

New Features

  - Added Docker support with automated image builds and publishing to
    GitHub Container Registry (ghcr.io/iobr/iobr). The image is based on
    rocker/tidyverse and includes IOBR with all dependencies
    pre-installed (#Docker).
  - Added Conda/Mamba installation instructions to README for easier
    environment setup.
  - Expanded iobr_cor_plot() to support multi-group (3+) comparisons
    using Kruskal-Wallis test, in addition to the existing two-group
    Wilcoxon test (#28).

Bug Fixes

  - Fixed generateRef_DEseq2() to handle sparse single-cell RNA-seq data
    by using type = "poscounts" for size factor estimation, preventing
    errors when all genes contain at least one zero.
  - Fixed batch_kruskal() output format to correctly display
    mean-centered values for all groups.
  - Fixed count2tpm() computation results to align with previous correct
    versions.
  - Updated count2tpm() documentation to clarify that gene identifiers
    are converted to gene symbols in the output regardless of input ID
    type.
  - Wrapped some long-running examples in \dontrun{} and \donttest{} to
    comply with CRAN check time limits.

Improvements

  - Refactored internal helper functions in count2tpm.R for better code
    organization and maintainability.
  - Updated load_data() documentation to include additional available
    datasets.

                        Changes in version 2.1.0                        

  - Initial CRAN submission.