NEWS

---

SlimR 1.1.5 (2026-06-03)

• New Features

  • Added Celltype_Compare(): a robust function for cross‑tabulating cell type labels and grouping variables between two single‑cell objects (e.g., Seurat). It automatically aligns cell barcodes using multiple normalisation strategies, returns count tables, column‑wise proportion tables, a dominant mapping, and a publication‑ready heatmap.

• Improvements

  • Updated general documentation and README structure.

SlimR 1.1.4

• New Features

  • Added a ASCII startup banner that displays when the package is attached.

• Improvements

  • Optimized citation output formatting and reduced duplication in .onAttach.

SlimR 1.1.3 (2026-04-07)

• Bug Fixes
  • AUCell scoring now uses full-transcriptome ranking. Previously, genes were ranked only among marker genes rather than the entire transcriptome, producing biologically incorrect percentile thresholds. The corrected implementation ranks all genes in the assay for each cell (chunked for memory efficiency), consistent with the original AUCell methodology.
  • Removed improper library() calls from Celltype_Calculate() to comply with R packaging best practices. All dependencies are now accessed exclusively through NAMESPACE imports, preventing namespace pollution.
  • Fixed ggplot2 deprecation warning: replaced geom_line(size = ...) with geom_line(linewidth = ...) in ROC curve plotting.

SlimR 1.1.2 (2026-03-13)

• New Features
  • Integrated the ScType marker database for tissue-aware cell type annotation.
  • Added Markers_filter_ScType() for filtering ScType markers by tissue type and cell name.
  • Added ScType, ScType_raw, and ScType_table data objects.

SlimR 1.1.1 (2026-02-05)

This release introduces a major new feature for fine-grained cell type identification alongside significant improvements to accuracy, performance, and usability.

• New Feature: Per-Cell Annotation System

  • Added a new workflow for annotating individual cells, complementing the existing cluster-based approach.
  • Offers three scoring methods: weighted (default), mean, and AUCell.
  • Includes optional UMAP-based spatial smoothing to improve annotation consistency.

• Major Improvements to Per-Cell Annotation

  • Introduced an adaptive min_score = "auto" threshold that scales with the number of cell types, preventing excessive "Unassigned" labels in large marker sets (e.g., 30+ types).
  • Added a new min_confidence parameter for ratio-based filtering, providing more robust cell type discrimination than simple score differences.
  • Enhanced the weighted scoring method with new marker specificity and IDF-like weights.
  • Improved the AUCell method with adaptive n_top calculation and a combined binary/rank-weighted scoring strategy.
  • Annotation results now include a Raw_score_matrix and a Parameters list for better reproducibility and debugging.

• Performance & Stability

  • Optimized for large datasets with vectorized operations and memory-efficient processing.
  • Integrated the RANN package for optional 10-100x faster k-NN computations in spatial smoothing.
  • Added comprehensive validation and error handling in core functions like compute_adaptive_parameters().

• Testing & Documentation

  • Added a comprehensive test suite with 147 tests covering core functionality, NA handling, and workflows.
  • Reorganized documentation and README with clearer distinctions between cluster-based and per-cell annotation.

SlimR 1.1.0 (2026-01-20)

• Improvements
  • Optimized AUC calculation in Celltype_Calculate() to use individual gene AUCs for more robust predictions.
  • Enhanced the adaptive machine learning algorithm in Parameter_Calculate() for better model generalization.
  • Extended Parameter_Calculate() to include threshold parameter prediction.
  • Updated general documentation and README structure.

SlimR 1.0.9 (2025-12-17)

• New Features

  • Integrated two new pan-cancer immune cell reference databases:
    • PCTIT: Pan-cancer T cell markers.
    • PCTAM: Pan-cancer macrophage markers.

• Improvements

  • Added a has_colnames parameter to Read_excel_markers() to support reading Excel files without column headers.
  • Updated documentation.

SlimR 1.0.8 (2025-10-08)

• New Features

  • Implemented machine learning-based parameter recognition using Random Forest, Gradient Boosting, SVM, and an Ensemble learner.

• Improvements

  • Optimized data filtering for the Markers_list_scIBD database.
  • Enhanced FSS (Fraction of Samples Significant) calculation in Read_seurat_markers() for outputs from the presto package.
  • Improved console output formatting in Celltype_Verification().

SlimR 1.0.7 (2025-08-18)

• New Features

  • Added the Celltype_Verification() function for generating validation dot plots.
  • Introduced custom color parameters (colour_low, colour_high) for all plotting functions.

• Improvements

  • Enhanced Read_seurat_markers() with better compatibility for FindMarkers results from the presto package.
  • Standardized function names across the package for consistency.
  • Improved the internal messaging system for clearer user feedback.
  • General updates for CRAN compliance.

• Bug Fixes

  • Resolved various user-reported issues.

SlimR 1.0.6

• New Features

  • Integrated the scIBD human intestine cell reference database.
  • Added AUC (Area Under the Curve) calculation and visualization to the Celltype_Calculate() function.
  • Implemented AUC-based prediction correction.

• Improvements

  • Streamlined the formatting of function outputs.
  • General updates for CRAN compliance.

• Bug Fixes

  • Fixed critical bugs in the core prediction pipeline.

SlimR 1.0.5

• New Features

  • Added the TCellSI T-cell reference database.
  • Introduced the Celltype_Calculate() function for automated cluster scoring.
  • Introduced the Celltype_Annotation() function for an end-to-end annotation workflow.

• Improvements

  • Enhanced the console message output system.
  • General updates for CRAN compliance.

• Bug Fixes

  • Resolved multiple code errors.

SlimR 1.0.4

• Improvements
  • Optimized the performance of the Celltype_annotation_Heatmap() function.
  • Enhanced the probability calculation in the helper function calculate_probability().
  • Changed the package license from GPL-3 to MIT.
  • General updates for CRAN compliance.

SlimR 1.0.3 (2025-07-29)

• Improvements
  • Updated Celltype_annotation_Heatmap() to use the new calculate_probability() function.
  • General updates for CRAN compliance.

SlimR 1.0.1

• Changes
  • Renamed Celltype_annotation_Bar() to Celltype_annotation_Box() and improved its visualization output.

SlimR 1.0.0

• Initial release on CRAN.
• Provides the core framework for cluster-based cell type annotation and visualization.