Changes in version 1.0.4                        

API Compatibility

  - Full compatibility with both Seurat v4 (4.0.0+) and Seurat v5
    (5.0.0+)
  - Added internal helper functions for version-aware API calls:
      - .getAssayData(): Auto-detects slot vs layer parameter
      - .getGraphs(): Compatible graph data access
      - .getMisc() / .setMisc(): Compatible misc slot access
      - .createAssayObject(): Compatible assay creation
  - Updated all vignettes to use Seurat-version-aware code patterns

                        Changes in version 1.0.3                        

Bug Fixes

  - Fixed p-value calculation to use proper two-tailed test (multiplied
    by 2)
  - Improved numerical stability in sparse.cor() function with handling
    of zero-variance columns
  - Fixed division-by-zero protection in C++ code for zero standard
    deviation columns
  - Fixed parameter name conflict (imputation → do_imputation) to avoid
    collision with function name
  - Fixed tag parameter handling when NULL in binomial family
  - Fixed loop variable scope issues in main scPAS function

Improvements

  - Added parallel computing support via n_cores parameter
  - Enhanced input validation with informative error messages
  - Improved sparse matrix operations in sparse_row_scale()
  - Cleaned up code comments and naming conventions
  - Added comprehensive test suite (57 tests)

Documentation

  - New vignettes:
      - Quick Start Guide (quick-start.Rmd)
      - Algorithm and Methodology (algorithm.Rmd)
      - Visualization Gallery (visualization.Rmd)
      - Case Study: Cancer Survival Analysis (case-survival.Rmd)
      - Case Study: Binary Classification (case-binary.Rmd)
  - Added pkgdown website support
  - Updated README with detailed usage examples

                        Changes in version 1.0.2                        

Bug Fixes

  - Fixed sparse matrix transpose issues in sparse.cor() function
  - Fixed rowMeans() and colMeans() handling for sparse matrices
  - Fixed FindNeighbors() rownames requirement
  - Improved logical indexing to avoid which() errors with sparse
    matrices
  - Enhanced NA handling in correlation calculations

Improvements

  - Added parallel computing support for permutation tests
  - Improved documentation and examples

                        Changes in version 1.0.1                        

Bug Fixes

  - Fixed cross-platform compatibility issues (Windows, macOS, Linux)
  - Fixed DLL/SO loading on different platforms

                        Changes in version 1.0.0                        

  - Initial release
  - Network-regularized sparse regression for phenotype-associated cell
    identification
  - Support for Gaussian, binomial, and Cox regression families
  - Integration with Seurat v4
  - KNN and ALRA imputation methods