Changes in version 0.99.8                        

Bug fixes

  - Fixed whole-blood examples in the introductory vignette and README
    to use the supplied normalized expression values directly, avoiding
    redundant CPM normalization and log2 transformation that caused the
    pkgdown build to fail during CLAMPfull().

                       Changes in version 0.99.7                        

Improvements

  - Added log2_transform = TRUE to preprocessCLAMP(),
    preprocessCLAMPFBM(), and cleanFBM(). Set it to FALSE to skip log2
    transformation while retaining missing-value handling and filtering.

Bug fixes

  - Changed preprocessCLAMP() to use population variance, matching
    preprocessCLAMPFBM() for variance filtering and downstream scaling.

                       Changes in version 0.99.6                        

Bug fixes

  - Corrected the pathway dot-plot heading and Visualization examples in
    the introductory vignette so all six plots are rendered.

                       Changes in version 0.99.5                        

Improvements

  - Added automated BiocCheck and cross-platform R CMD check workflows.
  - Added automatic pkgdown deployment after updates to devel.

                       Changes in version 0.99.4                        

Bug fixes

  - Made preprocessCLAMP() conditionally log2-transform and replace
    missing values before filtering, matching preprocessCLAMPFBM().
  - Made CLAMPplotU() handle missing FDR values when no pathways pass
    the requested thresholds.

                       Changes in version 0.99.0                        

New features

  - Initial Bioconductor submission.
  - Implements prior-informed latent variable decomposition for gene
    expression.
  - Supports file-backed matrices (bigstatsr::FBM) for large datasets.
  - Added CPM, z-score, and filtering functions for preprocessing.
  - Projection of new datasets into pre-trained models.
  - Vignettes with example workflows and detailed documentation.

Improvements

  - Optimized in-place operations for large matrices.
  - Added cross-validation utilities and adaptive sparsity support.
  - Roxygen2 documentation for all exported functions.

Bug fixes

  - Fixed edge case in CPM normalization for zero-count columns.
  - Corrected NA handling in preprocessing.